Kind: captions
Language: en
so today we have nadir bin ski he's a
professor at Northeastern University
working on various aspects of
computational agents that exhibit human
level intelligence
please give Nate a warm welcome thanks a
lot and thanks for having me here so the
title that was on the page was cognitive
modeling I'll kind of get there but I
wanted to put it in context so the the
bigger theme here is I want to talk
about what's called cognitive
architecture and if you've never heard
about that before that's great and I
wanted to contextualize that as how are
we what how is that one approach to get
us to AGI
and I say what my view of AGI is and put
up a whole bunch of TV and movie
characters that I grew up with that
inspire me that will lead us into what
is this thing called cognitive
architecture it's a whole research field
that crosses neuroscience psychology
cognitive science and all the way into
AI so I'll try to give you kind of the
historical big-picture view of it what
some of the actual systems are out there
that might be of interest to you and
then we'll kind of zoom in on one of
them that I've done a good amount of
work with called soar and what I'll try
to do is tell a story a research story
of how we started with kind of a core
research question we look to how humans
operate understood that phenomenon and
then took it and so really interesting
results from it and so at the end if
this field is of interest there's a few
pointers for you to go read more and and
go experience more of cognitive
architecture so just rough definition of
AGI given this in AGI class depending
the direction that you're coming from it
might be kind of understanding
intelligence or maybe developing
intelligent systems they're operating at
the level of human level intelligence
the the typical differences between this
and other sorts of maybe AI machine
learning systems we want systems that
are going to persist for a long period
of time
we want them robust to different
conditions we want them learning over
time and here's the crux of it working
on different tasks and in a lot of cases
tasks they didn't know we're coming
ahead of time I got into this because I
clearly watched too much TV and too many
movies and then I looked back at this
and I realized I think I'm covering 70's
80's 90's nots I guess it is and today
and so this is what I wanted out of AI
and this is what I wanted to work with
and then there's the the reality that we
have today
so instead of so who's watched Knight
Rider for instance I I don't think that
exists yet but but maybe we're getting
there
and in particular for fun during the
Amazon sale day I got myself an Alexa
and I could just see myself at some
point saying Alexa please might write me
an R sync script you know to sync my
class and if you have an Alexa you
probably know the following phrase this
this just always hurts me inside which
is sorry I don't know that one which is
okay right that's a lot of people have
no idea what I'm asking let alone how to
do that so what I want Alexa to respond
with after that is do you have time to
teach me and to provide some sort of
interface by which back and forth we can
kind of talk through this that we aren't
there yet to say the least but I'll talk
later about some work on a system called
Rosie that's working in that direction
we're starting to see see some ideas
about being able to teach systems out of
work
so folks who are in this field I think
generally fall into these three
categories they're just curious they
want to learn new things generate
knowledge work on hard problems great I
think there are folks who are in kind of
that middle cognitive modeling realm and
so I'll use this term a lot it's really
understanding how humans think how
humans operate human intelligence at
multiple levels and if you can do that
one there's just knowledge in and of
itself of how we operate but there's a
lot of really important applications
that you can think of if we were able to
not only understand but predict how
humans would respond react in various
tasks medicine is is an easy one there's
some work in HCI or HR I I'll get to
later where if you can predict how
humans would respond to a test you can
iterate tightly and develop better
interfaces it's already being used in
the realm of simulation and in defense
industries I happen to fall into the
latter group which or the bottom group
which is systems development which is to
say just the desire to build systems for
various tasks that are working on tasks
that kind of current AI machine learning
can't operate on and I think when you're
working at this level or on any system
that nobody's really achieved before
what do you do you you kind of look to
the examples that you have which in this
case that we know of
it's just humans right irrespective of
your motivation when you have kind of an
intent that you want to achieve in your
research you kind of let that drive your
approach and so I often show my AI
students this
the touring test you might have heard of
or variants of it that have come before
these were folks who are trying to
create system that acted in a certain
way that acted intelligently and the
kind of line that they drew the
benchmark that they used was to say
let's make systems that operate like
humans do
cognitive modelers will fit up into this
top point here to say it's not enough to
act that way but by some definition of
thinking we want the system to do what
humans do or at least be able to make
predictions about it so that might be
things like what errors would the human
make on this task or how long would it
take them to perform this task or what
emotion would be produced in this task
there are folks who are still thinking
about how the computer is operating but
it's trying to apply kind of rational
rules to it so a logician for instance
would say if you have a and you have B
the a gives you B B gives you see a
should definitely give you C that's just
what's rational and so there folks
operate in that direction and then if
you go to intro AI class anywhere in the
country particularly Berkeley because
they have graphics designers that I get
to steal from the benchmark would be
what the system produces in terms of
action and the benchmark is some sort of
optimal rational bound irrespective of
where you work in the space there's kind
of a common output that arrives when you
research these areas which is you can
learn individual bits and pieces and it
can be hard to bring them together to
build a system that either predicts or
acts on different tasks so this is part
of the transfer learning problem but
it's also part of having distinct
theories that are hard to combine
together so I'm going to give an example
that come comes out of cognitive
modeling or perhaps three examples so if
you were in a HCI class or some interest
psychology classes one of the first
things you'll learn about is Fitz law
which provides you the ability to
predict the difficulty level of
basically a human pointing
from where they start to a particular
place and it turns out that you can
learn some parameters and model this
based upon just the distance from where
you are to the targets and the size of
the target so both moving along distance
will take a while but also if you're
aiming for a very small point that can
take longer then if there's a large area
that you just kind of have to get
yourself to and so this is held true for
many humans so let's say we've learned
this and then we move on to the next
task
and we learn about what's called the
power law of practice which has been
shown true in a number of different
tasks what I'm showing here is one of
them where you're going to draw a line
through sequential set of circles here
starting at 1 going to 2 and so forth
not making a mistake or at least not
trying to and try to do this as fast as
possible and so for a particular person
we would fit the a B and C parameters
and we'd see a power law so as you
perform this task more you're going to
see a decrease in the amount of reaction
time required to complete the task great
we've learned two things about humans
let's add some more in so for those who
might have done some reinforcement
learning TV learning is one of those
approaches temporal difference learning
that's had some evidence of similar
sorts of processes in the dopamine
centers of the brain and it basically
says in a sequential learning tasks you
perform the task you get some sort of
reward how are you going to kind of
update your representation of what to do
in the future such as to maximize
expectation of future reward and there
are various models of how that changes
over time and you can build up functions
that allow you to form better and better
and better a given trial and error great
so we've learned three interesting
models here that hold true over multiple
people multiple tasks and so my question
is if we take these together and add
them together how do we start to
understand a task as quote/unquote
simple as chess which is to say we could
ask questions how how long would it take
for a person to play what mistakes would
they make
they played a few games how would they
adapt themselves or if we want to
develop system that ended up being good
at chess or at least learning to become
better at chess
my question is if you could there
doesn't seem to be a clear way to take
these very very individual theories and
kind of smash them together and get a
reasonable answer of how to play chess
or how do humans play chess and so
gentlemen in this slide is Alan Newell
one of the founders of AI did incredible
work in psychology and other fields
he gave a series of lectures at Harvard
in 1987 and they were published in 1990
called the unified theories of cognition
and his argument to the psychology
community at that point was the argument
on the prior slide they had many
individual studies many individual
results and so the question was how do
you bring them together to gain this
overall theory how do you make forward
progress and so his proposal was unified
theories of cognition which became known
as cognitive architecture which is to
say to bring together your core
assumptions your core beliefs of what
are the fixed mechanisms and processes
that intelligent agents would use across
tasks so the representations the
learning mechanisms the memory systems
bring them together implement them in a
theory and use that across tasks and the
core idea is that when you actually have
to implement this and see how it's going
to work across different tasks the
interconnections between these different
processes and representations would add
constraint and over time the constraints
would start limiting the design space of
what is necessary and what is possible
in terms of building intelligent systems
and so the overall goal from there was
to understand and exhibit human level
intelligence using these cognitive
architectures a'nature
question asked is okay so we've gone
from a methodology of science that we
understand how to operate in we make a
hypothesis we construct a study we
gather our data we evaluate that data
and we falsify we do not falsify the
original hypothesis and we can do that
over and over again and we know that
we're making for progress scientifically
if I've now taken that model and changed
it into I have a piece of software and
it's representing my theories and to
some extent I can configure that
software in different ways to work on
different tasks how do I know that I'm
making progress and so there's a form of
science called lactose ium and it's kind
of shown pictorially here where you
start with your core of what your your
beliefs are about where your head what
is necessary for achieving the goal that
you have and around that you'll have
kind of ephemeral hypotheses and
assumptions that over time may grow and
shrink and so you're trying out
different things trying out different
things and if an assumption is around
there long enough it becomes part of
that core and so as you work on more
tasks can learn more
either by your work or by data coming in
from with someone else the core is
growing larger and larger
you've got more constraints and you've
made more progress and so what I wanted
to look at we're in this community what
are some of the core assumptions that
are driving forward scientific progress
so one of them actually came out of
those lectures they're referred to as
Newell's time scales of human action and
so off on the left the left two columns
are both time units just expect somewhat
differently second from the left being
maybe more useful to a lot of us in
understanding daily life one step over
from there would be kind of at what
level processes are occurring so the
lowest three are down at kind of the
substrate the neuronal level we're
building up to deliberate tasks that
occur in the brain and tasks that are
operating on the order of ten seconds
some of these might occur in the
psychology laboratory but probably a
step up to
and ours and then above that really
becomes interactions between agents over
time and so if we start with that the
things to take away is that regular the
hypothesis is that regularities will
occur at these different time scales and
that they're useful and so those who
operate at that lowest time scale might
be considering neuroscience cognitive
neuroscience when you shift up to the
next couple levels what we would think
about in terms of the areas of science
that deal with that would be psychology
and cognitive science and then we shift
up a level and we're talking about
sociology and economics and the
interplay between agents over time and
so what we'll find with cognitive
architecture is that most of them will
tend to sit at the deliberate act we're
trying to take knowledge of a situation
and make a single decision and then
sequences of decisions over time will
build to tasks and tasks over time will
build to more interesting phenomenon I'm
actually going to show that that isn't
strictly true that there are folks
working in this field that actually do
operate one level below some other
assumptions so this is herb Simon
receiving the Nobel Prize in Economics
and part of what he received that award
for was an idea of bounded rationality
so in various fields we tend to model
humans as rational and his argument was
let's consider that human beings are
operating under various kinds of
constraints and so to model the rational
with respect to and bounded by how
complex the problem is that they're
working on how big is that search space
that they have to conquer cognitive
limitations so speed of operations
amount of memory short-term as well as
long-term as well as other aspects of
our computing infrastructure that are
going to keep us from being able to
arbitrarily solve complex problems as
well as how much time is available to
make that decision and so this is
actually a phrase that came out of his
speech when he received the Nobel Prize
decision-makers can satisfice either by
finding optimum
solutions for a simplified world just to
say take your big problem simplify in
some way and then solve that or by
finding satisfactory solutions for a
more realistic world take the world and
all its complexity take the problem in
all its complexity and try to find
something that works neither approach in
general dominates the other and both
have continued to co-exist and so what
you're actually going to see throughout
the cognitive architecture community is
this understanding that some problems
you're not going to be able to get an
optimal solution to if you consider for
instance bounded amount of computation
bounded time the need to be reactive to
a changing environment these sorts of
issues and so in some sense we can
decompose problems that come up over and
over again into simpler problems solve
those newer optimally or optimally fix
those in optimize those but more general
problems we might have to satisfy some
there's also the idea of the simple
system hypothesis so this is Alan Newell
and herb Simon they're considering how a
computer could play the game of chess so
the physical system physical symbol
system talks about the idea of taking
something some signal abstractly
referred to as symbol combining them in
some ways to form expressions and then
having operations that produce new
expressions a weak interpretation of the
idea that symbol systems are necessary
and sufficient for intelligent systems a
very weak way of talking about it is the
claim that there's nothing unique about
the neuronal infrastructure that we have
but if we got the software right we
could implement it in the bits bytes Ram
and processor that make up modern
computers that's kind of the weakest way
to look at this that we can do it with
silicon and not carbon stronger way that
this used to be looked at was more of a
logical standpoint which is to say if we
can encode rules of logic these tend to
line up if we think intuitively of
planning and problem solving and if we
can just get that right and get enough
fat's in there and enough
in there that somehow intelligence well
that's what we need for intelligence and
eventually we can get to the point of
intelligence and that's what you need
for intelligence and that was a starting
point that lasted for a while I think by
now most folks in this field would agree
that that's necessary to be able to
operate logically but that there are
going to be representations and
processes that will benefit from non
symbolic representation so particularly
perceptual processing visual auditory
and processing things in a more kind of
standard machine learning sort of way as
well as kind of statistic taking
advantage of statistical representations
so we're getting closer to actually
looking at cognitive architectures I did
want to go back to the idea that
different researchers are coming with
different research foci foci and we'll
start off with kind of the lowest level
and understanding biological modeling so
Leiber and spawn both try to model
different degrees of low-level details
parameters firing rates connectivities
between different kind of levels of
neuronal representations they build that
up and then they tried to build tasks
above that layer but always being very
cautious about being true to human
biological processes at a layer above
there would be psychological modeling
which is to say trying to build systems
that are true in some sense to areas of
the brain interactions in the brain and
being able to predict errors that we
made timing that we produced by the
human mind and so there I'll talk a
little bit about Akhtar this final level
down here these are systems that are
focused mainly on producing functional
systems that exhibit really cool
artifacts and and solve really cool
problems and so I'll spend most of the
time talking about soar but I want to
point out a relative newcomer in the
game called Sigma
so to talk about spawn a little bit
we'll see if the sound works in here I'm
going to let the Creator take this one
or not see how the AV system likes this
[Music]
but of course if I wouldn't be pleased
with a pad of the microseconds and all
celebrated since we're engineering is
critical goal engineering allows you to
break down equation intense very precise
descriptions which we can test like
building actual models one probably do
recently is called the song model this
Moscow has the two and a half million
individual mountains isolated and evens
the model is a knife and the up the
canal is Bernard
so essentially you can see images of
numbers and that is something like a
progressive in the case of the discarded
into that seat but magic tide reproduces
style that
yeah so for instance it's easy to get
the on environment and actually
forgive separately and silently on
medical side we all know that we have
cognitive town concession we get over
and we can try that address accountant
spicy alienating process with these nice
models another potential area in fact
it's on artificial intelligence a lot of
working out visual Imperfects thanks to
donations that are exceeded it at one
pass pretty since plane test what's
special is fine is that it's like that
many different paths and this X we have
not found it might appear californee the
flow of information through different
parts of the model something I haven't
seen very well so provide a pointer at
the end he's got a really cool book
called how to build a brain and if you
google and you can google spun you can
find a toolkit where you can kind of
construct circuits that will approximate
functions that you're interested in
connect them together set certain
properties that you would want at a low
level and build them up and actually
work on tasks at the level of vision and
robotic actuation so that's a really
cool system as we move into
architectures that are sitting above
that biological level I wanted to give
you kind of an overall sense of what
they're going to look like what a
prototypical architecture is going to
look like so they're gonna have some
ability to have perception the
modalities typically are more digital
symbolic but they will depending on the
architecture be able to handle vision
audition and various sensory inputs
these will get represented in some sort
of short-term memory whatever the
state's representation for the
particular system is there it's typical
to have a representation of the
knowledge of what tasks can be performed
when they should be performed how they
should be controlled and so these are
typically both actions that take place
internally that manage the internal
state of the system and perform internal
computations but also about external
actuation and external might be
a digital system a game AI but it might
also be some sort of robotic actuation
in real world there's typically some
sort of mechanism by which to select
from the available actions in a
particular situation there's typically
some way to augment this procedural
information which is to say learn about
new actions possibly modify existing
ones there's typically some semblance of
what's called declarative memory so
whereas procedural at least in humans if
I asked you to describe how to ride a
bike you might be able to say get on the
seats and pedal but in terms of keeping
your balance there you'd have a pretty
hard time describing it declaratively so
that's kind of the procedural side the
implicit representation of knowledge
whereas declarative would include facts
geography
math but it also include experiences
that the agent has had a more episodic
representation of declarative memory and
they'll typically have some way of
learning this information on mending it
over time and then finally some way of
taking actions in the world and they'll
all have some sort of cycle which is
perception comes in knowledge that the
agent has is brought to bear on that an
action is selected knowledge that knows
to condition on that action will act
accordingly both with internal processes
as well as eventually to take action and
then rinse and repeat so when we talk
about in an AI system an agent in this
context that would be the fixed
representation which is whatever
architecture we're talking about plus
set of knowledge that is typically
specific to the task but might be more
general so oftentimes these systems
could incorporate a more general
knowledge base of facts of linguistic
facts of Geographic facts
let's take Wikipedia and let's just
stick it in the brain of the system
there'll be more tasks in general but
then also whatever it is that you're
doing right now how should you proceed
in that and then it's typical to see
this processing cycle and going back to
the prior assumption the idea is that
these primitive cycles allow for the
agent to be reactive to its environment
so if new things come in that has react
to if the Lions sitting over there I
better run and maybe not do my calculus
homework right so as long as this cycle
is going I'm reactive but at the same
time if multiple actions are taken over
time I'm able to get complex behavior
over the long term so this is the act
our cognitive architecture it has many
of the kind of core pieces that I talked
about before let's see if the mouse yes
mouse is useful up there so we have the
procedural model here a short-term
memory is going to be these buffers that
are on the outside the procedural memory
is encoded as what I call production
rules or if-then rules if is this is the
state of my short-term memory this is
what I think should happen as a result
you have a selection of the appropriate
rule to fire and an execution you're
seeing associated parts of the brain
being represented here cool thing that
has been done over time in the act our
community is to make predictions about
brain areas and then perform MRIs and
gather that data and correlate that data
so when you use the system you will get
predictions about things like timing of
operations errors that will occur
probabilities that something is learned
but you'll also get predictions about to
degree that they can kind of brain areas
that are going to line light up and if
you want to that's actively being
developed at Carnegie Mellon to the left
is John Anderson who developed this
cognitive architecture ooh 30 ish years
ago and until the last about five years
he was the primary researcher developer
behind it with Christian and then
recently he's decided to spend more time
on cognitive tutoring systems and so
Christian has become the primary
developer there is an annual akhtar
workshop there's a summer school where
if you're thinking about modeling a
particular task you can kind of bring
your task to them bring your data they
teach you how to use the system and try
to get that study going right there on
the spot to give you a sense of what
kinds of tasks this could be applied to
so this is a representative of a certain
class of tasks certainly not the only
one let's try this again
I think powerpoints going to want a
restart every time okay so we're getting
predictions about basically where the
eye is going to move what you're not
seeing is it's actually processing
things like text and colors and making
predictions about what to do and how to
represent the information and how to
process the graph as a whole
I had alluded to this earlier there's
work by Bonnie John very similar so
making predictions about how humans
would use computer interfaces and at the
time she got hired away by IBM and so
they wanted the ability to have software
that you can put in front of software
designers and when they think they have
a good interface press a button this
model of human cognition would try to
perform the tasks that have been told to
do and make predictions about how long
it would take and so you can have this
tight feedback loop from designers
saying here's how good your particular
interfaces so act are as a whole it's
very prevalent in this community I went
to their web page and counted up just
the papers that they knew about it was
over 1,100 papers over time if you're
interested in it the main distribution
is in Lisp but many people have used
this and wanted to apply it to systems
that need a little more processing power
so there's the NRL has a Java port of it
that they use in robotics the Air Force
Research Lab and Dayton has implemented
it in Erlang for a parallel processing
of large declare knowledge bases they're
trying to do service-oriented
architectures with it CUDA because they
want what it has to say they don't want
to wait around for it to have to figure
that stuff out
so that's the two minutes about Akhtar
Sigma is a relative newcomer and it's
developed out at the University of
Southern California by man named Paul
rosenbloom mmm mentioned a couple
minutes because he was one of the prime
developers of soar at Carnegie Mellon so
he knows a lot about how store works and
he's worked on it over the years and I
think originally I'm gonna speak for him
and he'll probably say I was wrong I
think originally it was kind of a mental
exercise of can i reproduce or using a
uniform substrate I'll talk about so in
a little bit it's thirty years of
research code if anybody is dealt with
research code it's thirty years of C and
C++ with dozens of graduate students
over time it's not pretty at all and and
theoretically it's got these boxes
sitting out here and so he reimplemented
the the core functionality of soar all
using factor graphs and message passing
algorithms under the hood he got to that
point and then said there's nothing
stopping me from going further and so
now it can do all sorts of modern
machine learning vision optimization
sort of things that would take some time
in any other architecture to be able to
integrate well so it's been an
interesting experience it's now going to
be the basis for the virtual human
project out at the Institute for
Creative Technology it's Institute
associated with University of Southern
California for him until recently could
get your hands on it but in the last
couple years he's done some tutorials on
it he's got a public release with
documentation so that's something
interesting to keep an eye on but I'm
gonna spend all the remaining time on
the Soraa cognitive architecture and so
you see it looks quite a bit like the
prototypical architecture and I'll give
a sense again about how this all
operates give a sense of the people
involved we already talked about Alan
Newell so both John Laird who is my
advisor and Paul Rosenbloom were
students of Alan Newell John's thesis
project was related to the chunking
mechanism and soar which learns new
rules based upon sub-goal reasoning so
he finished that I believe the year I
was born and so he's one of the few
researchers you'll find who's still
actively working on their thesis project
beyond that's about I think about ten
years ago he founded soar technology
which is company up in Ann Arbor
Michigan while it's called solar
technology it doesn't do exclusively
soar but that's a part of the portfolio
general intelligence system stuff a lot
of Defense Association so some notes of
what's going to make soar different from
the other other architectures that fall
into this kind of functional
architecture category a big thing is a
focus on efficiency so john wants to be
able to run soar on just about anything
we just got on the soar mailing list a
desire to run it on a real-time
processor and our answer while we had
never done it before was probably it'll
work every release there's timing tests
and we always what we what we look at is
in a bunch of different domains for a
bunch of different reasons that relate
to human processing there's this magic
number that comes out which is 50
milliseconds which is to say in terms of
responding to tasks if you're above that
time humans will sense a delay and you
don't want that to happen now if we're
working in a robotics task 50
milliseconds if you're dramatically
above that you just fell off the curb or
worse or you just hit somebody in a car
right so we're trying to keep that as
low as possible and for most agents it
it doesn't even register it's below 1
millisecond fractions of millisecond but
I'll come back to this because a lot of
the work that I was doing was computer
science AI and a lot of efficient
algorithms and data structures and 50
milliseconds was that very high upper
bound it's also one of the projects that
has a public distribution you can get in
all sorts of operating systems we use
something called swig that allows you to
interface with it in a bunch of
different languages we kind of describe
the meta description and you are able to
basically generate bindings and
different platforms Korres C++ there was
a team at sore tech that said we don't
like C++ it gets messy so they actually
did a port over to pure Java in case
that appeals to you there's an annual
soar workshop that takes place in Ann
Arbor typically it's free you can go
there get a sort tutorial and talk to
folks who are working on soar and it's
fun I've been there every year but one
in the last decade it's just fun to see
the people around the world that are
using the system and all sorts of
interesting ways to give you a sense of
the diversity of the applications one of
the first was our one store which was
back in the days when it was an actual
challenge to build a computer which is
to say that your choice of certain
components would have radical
implications for other parts of the
computer so it wasn't just the Dell
website where you just I want this much
RAM I want this much CPU there was a lot
of thinking that went behind it and then
physical labor that went to construct
your computer and so it was making that
process a lot better there are folks
that apply to natural language
processing
I saw r7 was the core of the virtual
humans project for a long time
HCI tasks terrasaur was one of the
largest rule-based systems
tens of thousands of rules over 48 hours
it was a very large-scale simulation a
defense simulation lots of games it's
been applied to for various reasons and
then in the last few years
porting it on to mobile robotics
platforms this is Edwin Olsen's splinter
bot an early version of it that went on
to win the magic competition then I went
on to put soar on the web and if after
this talk you're really interested in a
dice game that I'm going to talk about
you can actually go to the iOS App Store
and download it's called Michigan liar's
dice it's free you don't have to pay for
it but you can actually play a liar's
dice with soar and it's even set the
difficulty level it's pretty good it
beats me on a regular basis I wanted to
give you a couple other just kind of
really weird feeling sort of
applications and really cool
applications the first one
is out of Georgia Tech go PowerPoint is
dom-based
interactive art installation in which
she participants can engage and
collaborate the movement improvisation
with each other and virtual advance
permits this thing her actually creates
a hyperspace English virtual and quicker
real bodies meet the line between human
and non-human is learned through images
to examine a relationship with
technology the night installation
ultimately examines how humans and
machine can co-create experiences and it
ducks out in a playful environment the
don't creates a social space that
encourages human human interaction and
collective dance experiences allowing
the depends to create an explorer
movement while having fun the
development of lumini has been a hundred
exploration in our forms of theatre and
dance as well as research and artificial
intelligence and cognitive science
lumahai draws inspiration from the
ancient art form of shot here the
original two-dimensional version of the
installation led the conceptualization
of the dome in the liminal space which
even silhouettes and virtual character
is being danced together on the
projection surface rather than relying
on a predominant library of movement
responses the virtual dancer learns in
this part measurements and utilizes new
points movement theory to systematically
reason about them and working
improvisational shoes under the moon
response the points theory is based in
dance and theater and analyzes the
performance along the dimensions of
tempo duration repetition kinesthetic
response shape spatial relationships
gesture architecture and
Photography the virtual dancer is able
to use several different strategies to
respond to human movements
these include mimicry of a movement
transformation of the movement along
viewpoints and mentions we're calling a
similar or complementary movement from
memory in terms of you fight revolutions
and define actually sponsor patterns of
the agent has learned while dancing with
its human partner the reason we did this
is this is part of a larger effort in
our lab for understanding the
relationship between
compeition cognition and creativity
where a large amount of our efforts go
into understanding human creativity and
how we make things together out were
created together as a way that almost
understand how we can build co-created
AI that serves the same purpose where to
be a colleague and collaborate with us
and create things with us so Brian was a
graduate student in John leritz
lab as well before I start this I lude
into this earlier where we're getting
closer to rosie saying can you teach me
so let me give you some introduction to
this in the lower left you're seeing the
view of a Kinect camera onto a flat
surface there's a robotic arm mainly 3d
printed parts few servos above that
you're seeing an interpretation of the
scene we're giving it kind of
associations of the four areas with
semantic titles like one is the table
one is the garbage just just semantic
terms for areas but other than that the
agent doesn't actually know all that
much and it's going to operate in two
modalities one is we'll call it natural
language natural ich language restricted
subset of English as well as some quote
unquote pointing so you're gonna see
some Mouse pointers in the upper left
saying I'll talk about this and this is
just a way to indicate location and so
starting off we're gonna say things like
you know pick up the blue block and it's
gonna be like I don't know what
is what is blue we say oh well that's a
color okay
you know so go get the green thing
what's green oh it's a color okay move
the blue thing to a particular location
where's that point it okay what is
moving like really it has to start from
the beginning and it's described and it
said okay now you've finished and once
we got to that point now I can say move
the green thing over here and it's got
everything that it needs to be able to
then reproduce the task given new
parameters and it's learned that ability
so let me give it a little bit of time
so you can look a little bit at top left
in terms of the pointers you're going to
see some text commands being entered so
what kind of attribute is blue we're
gonna say it's a color and so that can
map it then to a particular sensory
modality this is green so the pointing
what kind of thing is green okay color
so now it knows how to understand blue
and green as colors with respect to the
visual scene move rectangle to the table
what is rectangle okay now I can map
that on to or understanding parts of the
world is this the blue rectangle so the
arm is actually pointing itself to get
confirmation from the instructor and
then we're trying to understand in
general when you say move something what
is the goal of this operation and so
then it also has a declared
representation of the idea of this task
not only that it completed it then it
can look back on having completed the
task and understand what were the steps
that led to achieving a particular goal
so in order
move it you're gonna have to pick it up
it knows which one the blue thing is
great now
in the table so that's a particular
location and at this point we can say
you're done you have accomplished the
moved blue rectangle to the table and so
I can understand what that very simple
kind of process is like and associate
that with the verb to move and now we
can say move the green object or not do
the garbage and without any further
interaction based upon everything that
learned up till that point it can
successfully complete that task so this
is work of chavala Mohan and others at
the shore group at the University of
Michigan on the bruisy project and
they're extending this to playing games
and learning the rules of games through
text-based descriptions and multimodal
experience so in order to build up to
here's a story and so I wanted to give
you a sense of how research occurs in
the group and so there's these back and
forth that occur over time between
there's this piece of software called
soar we want to make this thing better
and give it new capabilities and so all
our agents are going to become better
and we always have to keep in mind and
you'll see this as I go further that it
has to be useful to a wide variety of
agents it has to be task independent and
it has to be efficient for us to do
anything in the architecture all of
those have to hold true so we do
something cool in the architecture and
then we say okay let's solve a cool
problem so it's build some agents to do
this and so this ends up testing what
are the limitations what are the issues
that arise in a particular mechanism as
well as integration with others and we
get to solve interesting problems we
usually find there was something missing
and then we can go back to the
architecture and rinse and repeat just
to give you an idea again how sore works
so the working memory is actually a
directed connected graph the perception
is just a subset of that graph and so
there's going to be symbolic
representations of most of the world
there is a visual subsystem in which you
can provide a scene graph just not
showing it here actions are also a
subset of that graph and so the
procedural knowledge which is also
production rules can modify can
sections of the input modify sections of
the output as well as arbitrary parts of
the graph to take actions so the
decision procedure says of all the
things that I know to do and I've kind
of ranked them according to various
preferences what single things should I
do
semantic memory for facts there's
episodic memory the agent is always
actually storing every experience it's
ever had over time in episodic memory
and it has the ability to get back to
that and so the similar cycle we saw
before we get input in this perception
called the input link rules are going to
fire all in parallel and say here's
everything I know about the situation
here's all the things I could do
decision procedure says here's what
we're going to do based upon the
selected operator all sorts of things
could happen with respect to memories
providing input rules firing to perform
computations and as well as potentially
output in the world and remember agent
reactivity is required we want the
system to be able to react to things in
the world at a very quick pace so
anything that happens in this cycle at
max the overall cycle has to be under 50
milliseconds and so that's gonna be
constraint we hold ourselves to and so
the story I'll be telling will say how
we got to a point where we started
actually forgetting things and we're an
architecture that doesn't want to be
like humans we want to create cool
systems but what we realized was
something that we do there's probably
some benefit to it and we actually put
it into our system in the lead to good
outputs so here's the research path I'm
going to walk down we had just a simple
problem which was we have these memory
systems and sometimes they're going to
get a cue that could relate to multiple
memories and the question is if you have
a fixed mechanism what should you return
in a task independent way which one of
these many memories should you return
that was our question and we looked to
some human data on this something called
the rational analysis of memory done by
John Anderson and realized that in human
language there are recency and frequency
effects that maybe it would be useful
and so we actually did an analysis found
that not only
does this occur but it's useful in what
are called word sense disambiguation
tasks I'll get to that what that means
in a second develop some algorithms to
scale this really well and it turned out
to worked out well not only in the
original task when we learn look to two
other completely different ones
the same underlying mechanism ended up
producing some really interesting
outputs so let me talk about word sense
disambiguation real quick this is a core
problem in natural language processing
if you haven't heard of it before let's
say we have an agent and for some reason
it needs to understand the verb to run
looks to its memory and finds that it
could you know run in the park it could
be running a fever could run an election
it could run a program and the question
is what should an task independent
memory mechanism return if all you've
been given is the verb to run and so the
rational analysis of memory looks
through multiple text corpora and what
they found was if a particular word had
been used recently it's very likely to
be reused again and if it hadn't been
used recently
there's going to be this effect where
the expression here the T is time since
the most recent use it's going to sum
those with a exponential decay and so
what it looks like if time is going to
the right activation hire as better as
you get these individual usages you get
these little drops and then eventually
drop down and so if we had just one
usage of a word the read would be what
the decay would look like and so the
core problem here is if we're at a
particular point and we want to select
between kind of the blue thing or the
red thing blue would have a higher
activation and so maybe that's useful
this is how things are modeled with
human memory but is it useful in general
for tasks and so we looked at common
corpora used in word sense
disambiguation and just said well if we
just look at this corporate twice and we
just use answers prior answers you know
I ask the question what is the sense of
this word I took a guess I got the right
answer and I used that recency and
frequency information in my task
independent memory would that be useful
and somewhat of a surprise but somewhat
maybe not of a
it actually performed really well across
multiple corpora so we said okay this
seems like a reasonable mechanism let's
look at implementing this efficiently in
the architecture and the problem was
this term right here said for every
memory for every time step you're having
to pay everything that doesn't sound
like a recipe for efficiency if you're
talking about lots and lots of knowledge
over long periods of time so we made use
of a nice approximation that petrol that
come up with to approximate tale effect
so accesses that happen
long long ago we could basically
approximate their effect on the overall
sum so now we had a fixed set of values
and what we basically said is since
these are always decreasing and all we
care about is relative order let's just
only recompute when someone gets a new
value so it's a guess it's a heuristic
and approximation but we looked at how
this worked on the same set of corpora
and in terms of query time if we made
these approximations well under our 50
millisecond the effect on task
performance was negligible in fact hunt
a couple of these it got ever so
slightly better terms of accuracy and
actually if we looked at the individual
decisions that were being made making
these sorts of approximations were
leading to up to 90 sorry at least 90
percent of the decisions being made were
identical to having done the true full
calculation so I said this is great
and we implemented this and worked
really well and then we started working
on what seemed like completely unrelated
problems
one was in mobile robotics we had a
mobile robot I'll show picture of in a
little while roaming around the halls
performing all sorts of tasks and what
we're finding was if you have a system
that's remembering everything in your
short-term memory and your short-term
memory gets really really big I don't
know about you my short-term memory
feels really really small I would love
it to be big but if you make your memory
really big and you try to remember
something
you're not having to pull lots and lots
and lots of information into your
short-term memory so the system was
actually getting slower simply because
it had a lot of short-term memory
representation of the overall map it was
looking up so large working memory a
problem Liars dices game you play with
dice we were doing in our L base system
on this reinforcement learning and it
turned out it's a really really big
value function we're having to store
lots of data and we didn't know which
stuff we had to keep around to keep the
performance up so we had a hypothesis
that forgetting was actually going to be
a beneficial thing that maybe maybe the
problem we have with our memories that
we really really dislike this forgetting
thing maybe it's actually useful and so
we experimented with the following
policy we said let's forget a memory if
one we haven't really it's not predicted
to be useful by this base level
activation we haven't used it recently
we haven't used it frequently maybe it's
not worth it
that and we felt confident that we could
approximately reconstruct it if we
absolutely had to and if those two
things held we could forget something so
it's this bait same basic algorithm but
instead of the ranking them it's if we
set a threshold for base level
activation finding when it is that a
memory is going to pass that threshold
and try to forget based upon that in a
way that's efficient that isn't going to
scale really really poorly so we were
able to come up with an efficient way to
implement this using an approximation
that ended up for most memories to be
exactly correct to the original I'm
happy to go over details of this if
anybody's interested later but end up
being a fairly close approximation one
that as compared to an accurate
completely accurate search for the value
ended up being somewhere between 15 to
20 times faster and so when we looked at
our mobile robot here oh sorry let me
get this back because our little robots
actually going around
it's the third floor of the computer
science building at the University of
Michigan it's going around he's building
a map and again the idea was this map is
getting too big so here was the basic
idea as the robots going around it's
going to need this map information about
rooms the color there is describing kind
of the strength of the memory and as it
gets farther and farther away and it
hasn't used part of the map for planning
or other purposes basically make it 2 K
away so that by the time it gets to the
bottom it's forgotten about the top but
we had the belief that we could
reconstruct portion that map if
necessary and so the hypothesis was this
would take care of our speed problems
and so what we looked at was here's our
50 millisecond thresholds if we do no
forgetting whatsoever
bad things were happening over time so
just 3,600 seconds this isn't a very
long time we're passing that threshold
this is dangerous for the robot if we
implement a task specific basically
cleanup rules which is really hard to
get right
that basically solved the problem when
we looked at our general forgetting
mechanism that we're using in other
places at an appropriate level of decay
we were actually doing better than
hand-tuned rules so this was kind of a
surprise win for us the other task seems
totally unrelated it's a dice game you
cover your dice you make bids about what
are under other people's cups this is
played in Pirates of the Caribbean when
they're on the boat in the second movie
and bidding for lives of service
honestly this is a game we love to play
in the University of Michigan lab and so
we're like hmm could soar play this and
so we built a system that could learn to
play this game rather well with
reinforcement learning and so the basic
idea was in a particular state of the
game soar would have options of actions
to perform it could construct estimates
of their associated value it would
choose one of those and depending on the
outcome something good happened you
might update that value and the big
problem was that the size of the state
space the number of possible states and
actions just is enormous and so memory
was blowing up and so what we said
similar sort of hypothesis if we decay
away these estimates that we could
probably reconstructs and we haven't
used it in a while our things going to
get better and so if we don't forget it
all
40,000 games isn't a whole lot when it
comes to reinforcement learning we were
up at two gigs we wanted to put this on
an iPhone that wasn't going to work so
well there had been prior work that had
used a similar approach they were down
at four or five hundred Meg's the
iPhones are not going to be happy
but it'll work so that gave us some hope
and we implemented our system okay
we're somewhere in the middle we can fit
on the iPhone a very good iPhone maybe
an iPad the question was though one
efficiency yeah we we fit under our 50
milliseconds but - how does the system
actually perform when you start
forgetting stuff can it learn to play
well and so y-axis here you're seeing
competency you play a thousand games how
many do you win so the bottom here 500
that's you know flipping a coin whether
or not you're going to win if we do know
forgetting whatsoever this is a pretty
good system the prior work while keeping
the memory low is also suffering with
respect to how well it was playing the
game and kind of cool was the system
that was basically more than having the
memory requirement was still performing
at the level of no forgetting whatsoever
so just to bring back why I went through
this story was we had a problem we
looked to our example of human level AI
which is humans themselves
we took an idea it turned out to be
beneficial we found in efficient
implementations and then found it was
useful in other parts of the
architecture and other tasks that didn't
seem to relate whatsoever but if you
download soar right now you would gain
access to all these mechanisms for
whatever task you want it to perform
just to give some sense in the field of
cognitive architecture what some of the
open issues are I think this is true in
a lot of fields in AI but integration of
systems over time the goal was they
wouldn't have all these theories and so
you could just kind of build over time
particularly when folks are working on
different architectures that becomes
hard but also when you have very
different initial starting points that
can still be an issue transfer learning
is an issue we're building into the
space of multimodal representations
which is to say not only abstract
symbolic but also visual wouldn't it be
nice if we had auditory and other senses
but building that into memories and
processing is still an open question
there's folks working on metacognition
which is to say the agent self assessing
its own State its own processing some
work has been done in here but still a
lot and I think the last one is a really
important question for anybody taking
this kind of class which is what would
happen if we did succeed if we did make
human-level AI and if you don't know
that picture right there
it's from a show that I recommend that
you watch that's by the BBC it's called
humans and it's basically what if we
were able to develop what are called
synths in the show think the robot that
can clean up after your laundry and cook
and all that good stuff interact with
you it looks and interacts as a human
but is completely our servants and then
hilarity and complex issues ensue so I
highly recommend if you haven't seen
that to go watch that I think these days
there's a lot of attention play pay to
machine learning and particular deep
learning methods as well it should
they're doing absolutely amazing things
and often the question is well you're
doing this and there's deep learning
over there you know how do they compare
and I honestly don't feel that that's
always a fruitful question because most
of the time they tend to be working on
different problems if I'm trying to find
objects in the scene I'm gonna pull out
tensorflow I'm really not going to pull
outs or it doesn't make sense it's not
the right tool for the job they haven't
been said there are times when they tend
to work together really really well so
the Rosi system that you saw there there
was some I believe neural networks being
used in the object recognition
mechanisms for the vision system there's
TD learning going
in terms of the dice game where we can
pick and choose and use this stuff
absolutely great because there are
problems that are best solved by these
methods so why avoid it and then on the
other side if you're trying to develop a
system where you you know in different
situations know exactly what you want
the system to do soar or other rule
based systems end up being the right
tool for the right job
so absolutely why not make it a piece of
the overall system some recommended
readings and some venues I'd mentioned
unified theories of cognition this is
Harvard Press I believe the short
cognitive architecture was MIT press
came out in 2012
I'll say I'm co-author and theoretically
would get proceeds but I've donated them
all to the University of Michigan so I
can just make this recommendation free
of ethical concerns personally it's an
interesting book it brings together lots
of history and lots of the new features
it's if you're really interested in soar
it's an easy sell
I'd mentioned crystallize Smith's how to
build a brain really cool read download
the software go through toriel's it's
it's really great how can the human mind
occur in the physical universe is one of
the court akhtar books so it talks
through a lot of the psychological
underpinnings and how the architecture
works it's a fascinating read one of the
papers trying to remember what year 2008
this goes through a lot of different
architectures in the field it's ten
years old but it gives you a good kind
of broad sweep if you want something a
little more recent this is last month's
issue of AI magazine completely
dedicated to cognitive systems so it's a
good place to look for the sort of stuff
in terms of academic venues triple AI
often has cognitive systems track
there's a conference called aiccm
international conference on cognitive
modeling where you'll see kind of a span
from biologic all the way up to AI
cognitive science or cogs AI they have a
conference as well as a journal ACS has
a conference as well as an online
journal advances in cognitive systems
cognitive systems research is a journal
that has a lot of this good stuff
there's AGI the conference
Vica is biologically inspired cognitive
architectures and I had mentioned both
there's a soar workshop and an act our
workshop that go on annually so leave it
at this there's some contact information
there and a lot of what I do these days
actually involves kind of explainable
machine learning integrating that with
cognitive systems as well as
optimization and robotics that scales
really well and also integrates with
cognitive systems so thank you if you
have a question please line up to one of
these two microphones so what what are
the main heuristics that you're using in
soar there can be heuristics at the task
level in the agent level or there's the
heuristics that are built into the
architecture to operate efficiently so
I'll give you a core example that comes
into the architecture and it's a fun
trick that if you're a programmer you
could use all the time which is only
process changes which is to say one of
the cool things about soar is you can
load it up with literally billions of
rules and I say literally because we've
done it and we know that it can turnover
still in under a millisecond and this
happens because instead of most systems
which process all the rules we just say
well anytime anything changes in the
world that's what we're going to react
to and of course if you look at the
biological world similar sorts of tricks
are being used so that's one of the core
ones that actually permeates multiple of
the mechanisms when it comes to
individual tasks it really is task
specific what that is so for instance
with the liar's dice game if you were to
go and download it when you're setting
the level of difficulty of it what
you're basically selecting is the subset
of heuristics that are being applied and
it starts very simply with things like
if I see lots of sixes then I'm likely
to believe a high number of sixes exist
but if I don't they're probably not
there at all so it's a start
but any Bayesian wouldn't really buy
that argument so then you start tacking
on a little bit of probabilistic
calculation and then it tacks on some
history of prior actions of the agents
so it really just builds now the Rosi
system one of the cool things they're
doing is game learning and specifically
having the agent be able to accept by a
text like natural text heuristics about
how to play the game even when it's not
sure what to do so you at one point you
mentioned about like generating new
rules yeah so I'm wondering like how do
you do that's so true and I'm the first
thing that comes to my mind are local
search methods okay so one thing is you
can actually implement heuristic search
in rules in the system and that's
actually how the robot navigates itself
so it does heuristic search but at the
level of rules generating new rules the
chunking mechanism says the following if
it's the case that in order to solve a
problem you had to kind of sub goal and
do some other work and you figure out
how to solve all that work and you've
got a result then and I'm greatly
oversimplifying but if you ever were in
the same situation again why don't I
just memorize the solution for that same
situation so it basically learns over
all the sub processing that was done and
encodes the situation I was in as
conditions and the results that were
produced as action and that's the new
rule all right thank you yeah hi so deep
learning and neural networks you know it
looks as though there's a bit of an
impedance mismatch between your system
and those types of system because you've
got a fixed kind of memory architecture
and they've got the memory and the rules
all kind of mixed together into one
system but could you interface your
system or a saw like system with deep
learning by playing in deep learning
agents has rules in your system so you'd
have to have some local memory but is
that is there some reason you can't plug
in deep learning as a kind of a rule
like module so I'm going to answer this
you work on it is
that's the been any work on that oh it's
yeah so I'll answer at multiple levels
one is you are writing a system and you
want to use both of these things how do
you make them talk and there is an API
that you can interface with any
environment and any set of tools and if
deep learning is one of them great and
if so or is the other one cool you have
no problem and you can do that today and
we have done this numerous times in
terms of integration into the
architecture all we have to do is think
of a sub-problem
in which all over simplify this but
basically function approximation is
useful I'm seeing basically kind of the
fixed structure of input I'm getting
feedback as to the output and I want to
learn the mapping to that over time if
you can make that case then you
integrate it as a part of the module
great and we have learning mechanisms
that do some of that deep learning just
hasn't been used to my knowledge to
solve any of those subproblems there's
nothing keeping it from being one of
those particularly when it comes down to
the low-level visual part of things a
problem that arises so I'll say what
would actually make some of this
difficult and it's a general problem
called simple grounding so at the level
of what most have what happens mostly in
store it is symbols being manipulated in
the highly discrete way and so how do
you get yourself from pixels and
low-level non symbolic representations
to something that's stable and discrete
and can be manipulated and that is
absolutely an open question in that
community and and that will make things
hard so spawn actually has an
interesting answer to that and it has a
distributive representation and it
operates over distributed
representations in what might feel like
a symbolic way so they're kind of ahead
of us on that but they're they're
starting from a lower point and so they
dealt with some of these issues and they
have a pretty good answer to that and
that's how they're moving up and that's
also why I showed Sigma which is at its
low level it's message passing
algorithms it's implementing
things like slam and Sat solving and
other sorts of really really it can
implement those on very low level
primitives but higher up it can also be
doing what soar is doing so there's an
answer there as well okay thank you so
another way of doing it would be to
layer the system so have one system
pre-processing the the the sensory input
or post-processing their draft but the
other one that would be another way of
combining two system and that's actually
what's going on in the rosey system so
the detection of objects in the scene is
a just just software that somebody wrote
I don't believe it's a deep learning
specifically but like the color
detection out of it I think is an SVM if
I'm correct so easily could be deep
learning
thanks you mentioned like the importance
of forgetting
in order for memory issues but you said
you could only forget because you could
reconstruct and then curse how do you
when you said we can start you need to
know that it happened before so do you
just compress the data like do you
really forget it order okay so and I put
quotes up and I said you think you can
reconstruct it so we came up with
approximations of this and so let me try
to answer this very grounded when it
comes to the mobile robot and you had
rooms that you had been to before the
entire map in its entirety was being
constructed in the robots semantic
memory so here's fats this room is
connected this room which is connected
this room which connected this room so
we had those sorts of representations
that existed up in at semantic memory
the rules can only operates down on
anything that's in short-term memory so
basically we were removing things from
the short-term memory and as necessary
be able to reconstruct it from the
long-term you could end up in some
situations in which you had made a
change locally in short-term memory
didn't get a chance to get it up and it
actually happened to be forgotten away
so you weren't guaranteed but it was
good enough that the connectivity
survived the agent was able to perform
the exact same task and we gained some
benefit for the RL system the rule we
came up with was the initial estimates
in the valley
you system which is here's how good I
think that is that's based on the
heuristics I described earlier some
simple probabilistic calculations of
counting some stuff that's where that
number came from we computed before we
could compute it again
the only time we can't reconstruct it
completely is if it had seen a certain
number of updates over time it's such a
large state space there are so many
actions so many states that most of the
states were never being seen so most of
those could be exactly reproduced by the
agent just thinking about it a little
bit and there were only a tiny tiny I'm
gonna say under 1% of the estimate the
value system that ever got updates and
that's actually not inconsistent with a
lot of these kinds of problems that have
really really large state spaces so I
think the statement was something like
if we had ever updated it don't forget
it and you saw that was already reducing
more than half of the memory load we
could have something higher to say 10
times something like that and that would
say we could reconstruct almost all of
it the prior work that I referenced was
strictly saying if it falls below
threshold no matter how many times in an
update no matter how much information
was there and so what we're adding was
probably can reconstruct and that was
getting us the the balance between the
efficiency and the ability to forget so
just under 7 you say we can probably we
can show it means that you keep trying
that you used to know it and so if you
need to be constructed you will but it's
just you're gonna run it again in some
times on the fly if I get back into that
situation and I happen to forget it the
the system knew how to compute it the
first time it goes and looks at all the
hand and it just pretends it's in that
situation for the very very first time
reconstructs that value estimate again
you're on that work question okay so the
actual mechanism of forgetting is
fascinating so l STM's rnns
have mechanisms for learning what to
forget and what not to forget have you
has there been any exploration of
learning the forgetting process just
doing something complicated or
interesting with which parts to forget
or not the closest I will say was kind
of a metacognition project that's 10 or
15 years old at this point which was
what happens when soar gets into a place
where it actually knows that it learned
something that's harmful to it that's
that's leading to poor decisions and in
that case it was still a very rule-based
process but it wasn't learning to forget
he was actually learning to override its
prior knowledge which might be closer to
some of what we do when we know we have
a bad habit we don't have a way of
forgetting that habit but instead we can
try to learn something on top of that
that leads to better operation in the
future to my knowledge that's the only
work at least in soar that's been done
just sorry I find the topic really
fascinating what lessons do you think we
can draw from the fact that forgetting
it's ultimately your the action of
forgetting is driven by the fact you
want to improve performance but do you
think forgetting is essential for AGI
the act of forgetting for building
systems that operate in this world how
important is forgetting I can think of
easy answers to that so one might be if
we take the cognitive modeling approach
we know humans do forget and we know
regularities of how humans forget and so
whether or not the system itself forgets
it's at least has to model the fact that
the humans that's interacting with are
going to forget and so at least it has
to have that ability to model in order
to interact effectively because if it
assumes we always remember everything
and it can't operate well in that
environment I think we're going to have
a problem is true forgetting going to be
necessary that's interesting our our AGI
system is going to hold a grudge for all
eternity we might want them to forget
this early age when we were
forcing them to work in our laboratory I
think I know what you're trying to yeah
exactly
yeah exactly and how do we build such a
system yeah anyways go ahead so I have
two quick two quick questions and one is
would you be able to speculate on how
you can connect function approximator
such as deep networks you know to
symbols and the second question
completely different this is regarding
your action selection I know we didn't
speak much about that when you have
different theories in your knowledge
representation and you have an action
selection which has to make construct a
plan by reasoning about the different
theories and the different pieces of
knowledge that are now held within your
memory or anything like all your rules
what kind of algorithms do you use in
the action selection to come up with the
plan you know is there any concept of
differentiation of the symbols or you
know or grammars or admissible grammars
and things like that that you use in
action selection I'm actually gonna
answer the second question first and
then you're gonna have to probably
remind me of what the first one was when
I get to the end so the action selection
mechanism one of these core tenants I
said is it's got to get through this
cycle fast so everything that's really
really built in has to be really really
simple and so the decision procedure is
actually really really simple it says
the rules are gonna fire the rules are
going the production rules are gonna
fire and there's gonna be a subset of
them that will say something like here's
an operator that you could select -
these are carlos acceptable operator
preferences they're ones that going to
say well based upon the fact that you
said that that was acceptable I think
it's the best thing or the worst thing
or I think 50/50 chance I'm going to get
reward out of this there's actually a
fixed language of preferences that are
being asserted and actually a nice fixed
procedure by which if I have a set of
preferences to make a very quick and
clean decision so what's basically
happened is you've pushed the hard
questions of how to make complex
decisions about actions up to a higher
level
the low level architecture is always
given a set of
Jen's going to be able to make a
relatively quick decision and it gets
pushed into the knowledge of the agent
to construct a sequence of decisions
that over time is going to get to the
more interesting questions you're
talking about but how can you reason
that that sequence will take you to the
goal that you desire so people is there
any guarantee on that is that in general
across tasks no but people have for
instance implemented a star I was
mentioning as wouls
right yeah so I know given certain
properties about the search tack that
task that's being searched based upon
these rules given a finite search space
eventually it will get there and if I
have a good heuristic in there I know
certain properties about the optimality
so I can reason at that level in general
I think this comes back to the
assumption I made earlier about bounded
rationality to say parts of the
architecture of solving subproblems
optimally the general problems that it's
going to work on it's going to try its
best based upon the knowledge that it
has and that's about the end of
guarantees that you can typically make
in the architecture okay I think your
first question was speculate on
connecting symbol approach I mean
function approximate is you know you
know you know multiple layer function
approximate is like deep learning
networks to two symbols that you can
reason about at a higher level yeah I
think that's a great open space if I had
time this would be somebody I'll be
working on right now which is somewhere
before it basically said taking in a
scene and then detecting objects out of
that scene and using those as symbols
and reasoning about those over time I
think the spawn work is quite
interesting so the symbols that they're
operating on are actually a distributed
representation of the input space and
the closest I can get to this is if
you've seen a word Tyvek where you're
taking a language corpus and what you're
getting out of there is a vector number
that has certain properties but it's
also a vector you can operate on as a
unit so it has nice properties you can
operate with it on other vectors you
know that if I got the same word in the
same context I would get back to that
exact same vector so those are that's
the kind of representation that seems
like it's going to be able to bridge
that chasm where we can get from sensory
information to something that can be
operated on and reasoned about in this
sort of symbolic architecture and get us
from there from actual sensory
information I had a question what do you
think are the biggest strengths of the
cognitive architecture approach compared
to other approaches in artificial
intelligence and the flip side of that
what do you think are the biggest
shortcomings of cognitive architecture
with respect to us with respect to you
being humans yeah a human level like
like what needs to be like how come
cognitive architecture has not solved
AGI because we want job security that's
the answer we've totally solved it
already so strengths I think
conceptually is keeping an eye on the
ball which is if what you're looking at
is trying to make human-level AI I it's
hard it's challenging it's ambitious to
say that's the goal because for decades
we haven't done it it's extraordinarily
hard it it is less difficult in some
ways to constrain it yourself down to a
single problem that having been said I'm
not very good at making a car drive
itself in some ways that's a simpler
problem it's great at challenging it of
itself and it'll have great impact on
humanity it's a great problem to work on
human level AI is huge it's not even
well-defined as a problem and so
what's the strength here bravery
stupidity in the face of failure
resilience over time keeping alive this
idea of trying to reproduce a level of
human intelligence that's more general I
don't know if that's a very satisfactory
answer for you
downside home runs are fairly rare and
by home run I mean a system that finds
its way to the the general populace to
the marketplace I'd mentioned Bonnie
Johns specifically because you know this
is twenty thirty years of research and
then she found a way that actually makes
a whole lot of sense under direct
application so it was a lot of a lot of
years of basic research a lot of
researchers and then there was there was
the big win there what was this one oh
this was a bunny John was a researcher
this was using akhtar models of I gaze
and reaction and so forth to be able to
make predictions about how humans would
use user interfaces so those sorts of
outcomes are rare it it if you work in
AI one of the first things you learn
about is blocks world it's kind of in
the classic AI textbook I will tell you
I've worked on that problem at about
three different variants I've gone to
many conferences where presentations
have been made about blocks world which
is to say we're good progress is being
made but the way you end up thinking
about is it really really small
constrained problems ironically you you
have this big vision but in order to
make progress that ends up being on
moving blocks on a table and so it's
it's a big challenge I just think it'll
take a lot of time the I'll say the
other thing they haven't we haven't
really gotten to although I brought up
spawn and I brought up Sigma
an idea of how to scale this thing
something I like about deep learning is
just some extent with lots of asterisks
and 10,000 foot view it's kind of like
well we've gotten this far all right
let's just provided different inputs
different outputs and we'll have some
tricks on the middle and suddenly you
have you know end to end deep learning
of a bigger problem and a bigger problem
there's a way to see how this expands
given enough data given enough computing
and incremental advances when it comes
to soar it takes not only a big idea but
it takes a lot of software engineering
to integrate it there's a lot of
constraints built into it it slows it
down so something like Sigma is oh well
I can change a little bit of the
configuration of the graph I can use
variants on the algorithm boom it's
integrated I can experiment fairly
quickly so starting with that sort of
infrastructure does not give you the
constraint you kind of want with your
big picture vision of going towards
human level AI but in terms of being
able to be agile in your research it's
it's kind of incredible Izzie
thank you you'd mention that ideas such
as base level decay at these techniques
they were based their original
inspirations were based off of human
cognition and and because humans can't
remember everything so were there any
instances of the other way around where
some discovery in cognitive modeling
fueled it another discovery in cognitive
science so what one thing I'm gonna
point out and your question was based on
the decay with respect to human
cognition the study actually was let's
look at text and properties of text and
use that to then make predictions about
what must be true about human cognition
so John Anderson and the other
researchers looked at believe it was New
York Times articles
his Oh John Anderson's emails and I'm
trying to remember what the third I
think it was parents utterances with
their kids or something like this it was
actually looking at text corpora and the
words that were occurring in at varying
frequencies that that analysis that
rational analysis actually led to models
that got integrated within the act arc
architecture that then became validated
through multiple trials that then became
validated with respect to MRI scans and
is now being used to both do study back
with humans but also develop systems
that interact well with humans so I
think that in and of itself ends up
being an example it's a cheat but the
UAV the soar UAV system I believe is a
single robot that has multi multiple
agents running on it so where is this I
got it off your website ok but either
way your systems allow for multi agents
ok so my question is how are you
preventing them from converging with new
data and are you changing what they're
forgetting selectively as one of those
ways so I'll say yes you can have multi
agent source systems on a single system
on multiple systems there's not any real
strong theory that relates to
multi-agent systems so there's no real
constraint there that you can come up
with a protocol for them interacting
each one is going to have its own set of
memories set of knowledge there really
is no constraint on you being able to
communicate like you would if it were
any other system interacting with soar
so I don't really think I have a great
answer for it so that is to say if you
had goo
Theory's good algorithms about how
multi-agent systems work and how they
can bring knowledge together form a
fusion sort of way it might be something
that you could bring to a multi agent
source system but there's nothing really
there to help you there's no mechanisms
there really to help you do that any
better than you would otherwise and you
would have to kind of constraints of
your representations the process as to
what it has fixed in terms of its sort
of memory and its sort of processing
cycle thank you
[Applause]