15. Learning Near Misses; Felicity Conditions Video Lecture - Artificial

PROFESSOR PATRICK WINSTON:
You know, some
of you who for instance--
I don't know, Sonya,
Krishna, Shoshana--
some of you I can count on
being here every time.
Some of you show up
once in a while.
The ones of you who show up once
in a while happen to be
very lucky if you picked today,
because what we're
going to do today is I'm going
to tell you stuff that might
make a big difference
in your whole life.
Because I'm going to tell
you how you can
make yourself smarter.
No kidding.
And I'm also going to tell you
how you can package your ideas
so you'll be the one that's
picked instead
of some other slug.
So that's what we're
going to do today.
It's the most important lecture
of the semester.
The sleep lecture is only the
second most important.
This is the most important.
Now the vehicle that's going
to get us there is a
discussion about how it's
possible to learn in a way
that is a little reminiscent
of what we
talked about last time.
Because last time we learned
something very definite from a
small number of examples.
This takes it one step further
and shows how it's possible to
learn in a human-like way from
a single example in one shot.
So it's extremely different,
very different from everything
you've seen before.
Everything that involves
learning from thousands of
trials and gazillions of
examples and only learning a
little tiny bit, if anything,
from each of them.
This is going to
learn something
definite from every example.
So here's the classroom
example.
What's this?
It's an arch.
I know the architects are
complaining that it's not an
arch in architecture land.
It's a post and lintel
construction.
But for us today it's
going to be an arch.
Now if you were from Mars and
didn't know what an arch was,
I might present this to you and
you'd get a general idea
of some things that might be
factors, but you'd have no
idea what's really important.
So then I would say,
that's not an arch.
And you would learn something
very definite from that.
And then I would shove these
together and put this back on,
and I would say, that's
not an arch either.
And you'd learn something
very definite from that.
And then I could paint the top
one blue, and you'd learn
something very different
from that.
And how can that happen
is the question?
How can that happen in detail,
and what might it mean for
human learning and how you can
make yourself smarter?
And that's where we're
going to go.
All right?
So how can we make a program
that's a smart as a martian
about learning things
like that?
Well, if you were writing that
program, surely the first
thing you would do is you'd try
to get off the picture as
quickly as possible and into
symbol land where things are
clearer about what the
important parts are.
So you'd be presented with an
initial example that might
look like this.
We'll call that an example.
And it's more than
just an example.
It's the initial model.
That's the starting point.
And now we're going to couple
that with something that's not
actually an arch but looks a
whole lot like one, at least
on the descriptive level to
which we're about to go.
So here's something that's not
an arch, but its description
doesn't differ from that
of an arch very much.
In fact, if we were to draw this
out in a kind of network,
we would have a description
that looks like this, and
these relations would be
support relations.
And this would be drawn
out like so.
And the only difference
would be--
the only difference would be
that those support relations
that we had in the
initial model--
the example--
have disappeared down out here
in this configuration.
But since it's not very
different from the model,
we're going to call
this a near miss.
And now, you see, we've
abstracted away from all the
details that don't
matter to us.
Last time we talked about a
good representation having
certain qualities--
qualities like making the
right things explicit.
Well, this makes the structure
explicit, and it suppresses
information about blemishes
on the surface.
We don't care much about how
tall the objects are.
We don't think it matters
what they're made of.
So this is a representation that
satisfies the first of
the criteria from last time.
It makes the right
things explicit.
And by making the right things
explicit, it's exposing some
constraint here with
respect to what it
takes to be an arch.
And we see that if those support
relations are missing,
it's not an arch.
So we ought to be able to learn
something from that.
What we're going to do is we're
going to put these two
things together.
We're going to describe the
difference between the two.
And we're going to reach the
conclusion that since there's
only one difference--
one kind of difference with
two manifestations to
disappearing support relations,
we're going to
conclude that those support
relations are important.
And we're going to
turn them red
because they're so important.
And we're going to change the
name from "support" to "must
support."
So this is our new model.
This is an evolving model that
now is decorated with
information about what's
important.
So if you're going to match
something against this model,
it must be the case that those
support relations are there.
If it's not there-- if they're
not there, it's not an arch.
All right?
So we've learned something
definite
from a single example.
This is not 10,000 trials.
This is a teacher presenting
something to the student and
the student learning something
immediately in one step about
what's important in an arch.
So let's do it again.
That was so much fun.
Let's do this one.
Same as before except that now
when we describe this thing,
there are some additional
relations--
these relations, and those
are touch relations.
So now when we compare that--
is that an arch?
No.
It's a near miss.
When we compare that near miss
with our evolving model, we
see immediately that once again
there's exactly one
difference, two
manifestations, the touch relations.
So we can immediately conclude
that these touch relations are
interfering with our belief that
this could be an arch.
So what do we do with that?
We put those together
again and we build
ourselves a new model.
It's much like the old model.
It still has the imperatives
up here.
We have to have the
support relations.
But now down here-- and we draw
not signs through there--
these are must not
touch relations.
So now you can't match against
that model if those two side
supports are touching
each other.
So in just two steps, we've
learned two important things
about what has to be in place in
order for this thing to be
construed to be an arch.
So our martian is making
great progress.
But our martian isn't through,
because there's some more
things we might want
it to know about
the nature of arches.
For example, we might present
it with this one.
Well, that looks just like
our initial example.
It's an example just like
our initial example.
But this time the top has
been painted red.
And I'm still saying that
that's an arch.
So once again, there's only
one difference and that
difference is that in the
description of this object, we
have the additional information
that the color of
the top is red.
And we've been carrying along
without saying so, that the
color of the top in the evolving
model is white.
So now we know that the top
doesn't have to be white.
It can be either red or white.
So we'll put those
two together and
we'll get a new model.
And that new model this
time once again
will have three parts.
It will have the relations, an
imperative form that we've
been carrying along now, the
must support and the must not
touch, but now we're going to
turn that color relation
itself into an imperative.
And we're going to say that
the top has to be
either red or white.
So now, once again, in one step
we've learned something
definite about archness.
Two more steps.
Suppose now we present
it with this example.
It's an example.
And this time there's going
to be a little paint
added here as well.
This time we're going to have
the top painted blue like so.
So the description
will be like so.
And now we have to somehow put
that together with our
evolving model to make
a new model.
And there's some choices here.
And our choice depends somewhat
on the nature of the
world that we're working in.
So suppose we're working
in flag world.
There are only three colors--
red, white, and blue.
Now we've seen them all.
If we've seen them all, then
what we're going to do is
we're going to say that the
evolving model now is adjusted
yet again like so.
Oh-- but those are imperatives
still.
Let me carry that along.
At this time, this guy--
the color relation--
goes out here to anything
at all.
So we could have just not drawn
it at all, but then we
would have lost track of the
fact that we've actually
learned that anything
can be there.
So we're going to retain the
relation but have it point to
the "anything goes" marker.
Well, we're making great
progress and I said there's
just one more thing to go.
So let me compress that
into this area here.
What I'm going to add this time
is I'm going to say that
the example is like everything
you've seen before except that
the top is now one of those
kinds of child's bricks.
So you have a choice actually
about whether this
is an arch or not.
But if I say, yeah, it's still
an arch, then we'd add a
little something to
its description.
So this description would
look like this.
Same things that we've seen
before in terms of support,
but now we'd have a relation
that says that
this top is a wedge.
And over here--
something we've been carrying
along but not writing down--
this top is a block.
A brick, I guess in the
language of the day.
So if we say that it can be
either a wedge or a brick on
top, what do we do with that?
Once again, it depends on the
nature of representation, but
if we say that we have a
representation, that has a
hierarchy of parts.
So bricks and wedges are both
children's blocks and
children's box or toys.
Then we can think of drawing
in a little bit of that
hierarchy right here and
saying well, let's see.
Immediately above that we've
got the brick or wedge.
And a little bit above
that we've got block.
And a little bit above
that we've got toy.
And a little bit above that
we eventually get to
any physical object.
So what does it do in response
to that kind of situation?
You have the choice.
But what the program I'm
speaking of actually did was
to make a conservative
generalization up here just to
say that it's one
of those guys.
So once again it's learned
something definite.
Let me see.
Let me count the steps.
One, two, three, four, five.
And I just learned
four things.
So the generalization of a
color, it took two steps to
get all the way up
to "don't care."
So note how it contrasts with
anything you've seen in a
neural net.
Or anything you will see
downstream in some of the
other learning techniques that
we'll be talking about that
involve using thousands of
samples to learn what it is--
to learn whatever it is that
is intended to be learned.
Let me show you another
example of how these
heuristics can be put to work.
So there are two sets
of drawings.
We have the upper set
and the lower set.
And your task, you smart
humans working in vast
parallelism, your task is to
give me a description of the
top trains that distinguishes
and separates them from the
trains on the bottom.
You got it?
Nobody's got it?
Well, let me try one on you.
The top trains all have a short
car with a closed top.
So how is it possible that
a computer could have
figured that out?
It turns out that it figured
it out with much the same
apparatus that I've shown you
here in connection with the
arches, just deployed in a
somewhat different manner.
In this particular case, the
examples are presented one at
a time by a teacher
who's eager for
the student to learn.
In this case, the examples are
presented all at once and the
machine is expected to figure
out a description that
separates the two groups.
And here's how it works.
What you do is you start
with one of them.
But you have a lot of them.
You have some examples--
we'll call the examples on top
the "plus examples" and the
examples on the bottom the
"negative examples." So the
first thing that you do is you
pick one of the positive
examples to work with.
Anybody got any good guesses
about what we're
going to call that?
Yeah, you do.
We're going to call
that the seed.
It's just highly reminiscent of
what we did last time when
we were doing [? phonology ?]
but now at a much
different level.
We're going to pick one of those
guys to be the seed, and
then we're going to take these
heuristics and we're going to
search for one that loosens this
description so that it
covers more of the positives.
You see, if you have a seed that
is exactly a description
of a particular thing and you
insist that everything be just
like that, then nothing will
match except itself.
But you can use these heuristics
to expand the
coverage of the description, to
loosen it so that it covers
more of the positives.
So in your first step you might
cover, for example, that
group of objects.
Too bad for your side, you've
also in that particular case
included a negative example in
your description, but perhaps
in this next step beyond that
you'll get to the point where
you've eliminated all of those
negative examples and zeroed
in on all the positive
examples.
So how might a program be
constructed that would do that
sort of thing?
Well, think about the choices.
The first choice that you have
it is to pick a positive
example to be the seed.
And once you've picked a
particular example to be the
seed, then you can apply
heuristics, all of them that
you have, to make a new
description that may cover the
data better.
It may have more of the
positives and fewer of the
negatives than in your
previous step.
But this, if you have a lot of
heuristics, and these are a
lot of heuristics because
there's a lot of description
in that set of trains, there
are lots of possible things
that you could do with those
heuristics because you could
apply them anywhere.
So this tree is extremely
large.
So what do you do to keep
it under control?
Well, now you have answers
to questions like that by
knee-jerk, right?
The branching factor
is too big.
You want to keep a few
solutions going.
You have some way of measuring
how well you're doing so you
can use a beam search.
This piece here was originally
worked out by a friend of
mine, now, alas, deceased,
[? Rashad ?]
[? Malkowski ?]
when he was at the University
of Illinois.
And of course, he wasn't
interested in toy trains, he
was just interested in
soybean diseases.
And so this exact program was
used to build descriptions of
soybean diseases.
It turned out to be
better than the
plant pathology books.
We now have two ways of
deploying the same heuristics.
But my vocabulary is in need
of enrichment, because I'm
talking about "those"
heuristics.
And one of the nice things
that [? Malkowski ?]
did for me a long time ago is
give each of them a name.
So here are the names that
were developed by
[? Malkowski. ?]
What's happening here?
You're going from an original
model to an understanding--
some things are essential.
So he called this the "require
link" heuristic.
And here in the next step, we're
forbidding some things
from being there.
So [? Malkowski ?]
called that heuristic the
"forbid link" heuristic.
And in the next step, we're
saying it can be
either red or white.
So we have a set of colors
and we're extending it.
And over here in this heuristic,
going from red or
white to anything goes, that's
essentially forgetting about
color altogether, so we're going
to call that "drop link"
even though for reasons of
keeping track, we don't
actually get rid of it.
We just have it pointing to
the "anything" marker.
And finally, in this last step,
what we're doing with
this tree of categories is we're
climbing up it one step.
So he called that the "climb
tree" heuristic.
So now we have a vocabulary
of things we can do in the
learning process, and having
that vocabulary gives us power
over it, right?
Because those are names.
We can now say, well, what
you need here is
the "drop link" heuristic.
And what you need over there is
the "extend set" heuristic.
So now I want to back up
yet another time and
say, well, let's see.
When we were working with that
phonology stuff, all I did was
generalize.
Are we just generalizing here?
No.
We're both generalizing
and specializing.
So when I say that the links
over here that are developed
in our first step are
essential, this is a
specialization step.
And when I say they can't be--
they cannot be touch
relations, that's a
specialization step.
Because we're able to match
fewer and fewer things when we
say you can't have
touch relations.
But over here, when I go here
and say, well, it doesn't have
to be white.
It can also be red.
That's a generalization.
Now we can match more things.
And when I drop the link
altogether, that's a
generalization.
And when I climb the tree,
that's a generalization.
And that's why when I do this
notional picture of what
happens when [? Malkowski ?]
program does a tree search to
find a solution to the train
problem, they're both
specialization steps which
draw in the number of things
that can be matched, and
generalization steps that
make it broader.
So, let's see.
We've also got the notion
of near miss.
And we've got the notion
of example--
some of these things
are examples,
some are near misses.
We've got generalization
specialization.
Does one go with one or the
other, or are they all mixed
up in their relationship
to each other?
Can you generalize and
specialize with near misses?
What do you think?
You think--
you don't think so,
[INAUDIBLE]?
What do you think?
STUDENT: [INAUDIBLE]
specialization.
PROFESSOR PATRICK WINSTON:
[INAUDIBLE] lead to
specialization.
Let's see if that's right.
So we've got specialization
here, and that's a near miss.
We've got specialization here,
and that's a near miss.
We've got generalization here,
and that's an example.
And we've got generalization
here, and that's an example.
And we've got generalization
here, and that's an example.
So [INAUDIBLE] has got
that one nailed.
The examples always generalize,
and the near
misses always specialize.
So we've got apparatuses in
place that allow us to both
expand what we could match and
shrink what we could match.
So what has this got
to do anything?
Well, which one of these methods
is better, by the way?
This one--
this one requires a teacher
to organize everything up.
This one can handle
it in batch mode.
This one is the sort of thing
you would need to do with a
human because we don't
have much memory.
That one is the sort of thing
that a computer's good at
because it has lots of memory.
So which one's better?
Well, it depends on what
you're trying to do.
If you're trying to build a
machine that analyzes the
stock market, you might
want to go that way.
Or soybean diseases, or
any one of a variety
of practical problems.
If you're trying to model
people, then maybe this is a
way that deserves additional
merit.
How do you get all
that sorted out?
Well, one way to get it all
sorted out is to talk in terms
of what are sometimes called
"felicity conditions." So when
I talk about felicity
conditions, I'm talking about
a teacher and a student
and covenants that
hold between them.
So here's the teacher.
That's me.
And here's the student.
That's you.
And the objective of interaction
is to transform an
initial state of knowledge into
a new state of knowledge
so that the student is smarter
and able to make use of that
new knowledge to do things that
couldn't be done before
by the student.
So the student over here
has a learner.
And he has something that
uses what is learned.
And the teacher over
here has a style.
So if any learning is to take
place, one side has to know
something about the
other side.
For example, it's helpful if
the teacher understands the
initial state of the student.
And here's one way of
thinking about that.
You can think of what you know
as forming a kind of network.
So initially, you don't
know anything.
But as you learn, you start
developing quanta of knowledge.
And these quanta of knowledge
are all linked together by
prerequisite relationships that
might indicate how you
get from one quantum
to another.
So maybe you have generalization
links, maybe
you have specialization links,
maybe you have combination
links, but you can think of what
you know as forming this
kind of network.
Now your state of knowledge at
any particular time can then
be viewed as a kind of wavefront
in that space.
So if I, the teacher, know where
your wavefront is, can I
do a better job of teaching
you stuff?
Sure, for this reason.
Suppose you make a mistake,
m1, that depends on q1.
Way, way behind your
wavefront.
What do I do if I know
that you made a
mistake of that kind?
Oh, I just say, oh, you forgot
you need a semicolon after
that kind of statement.
I just remind you of something
that you certainly know, you
just overlooked.
Right?
On the other hand, suppose you
make a mistake that depends on
a piece of knowledge
way out here.
That kind of mistake, m2.
What do I say to you then?
What do you think, Patrick?
What do you think I would
say if you made
that kind of mistake?
STUDENT: [INAUDIBLE].
PROFESSOR PATRICK WINSTON: No.
That's not what I would
say [INAUDIBLE].
STUDENT: You'd tell us that
we don't know that yet.
PROFESSOR PATRICK WINSTON: I
would say something like that.
What [INAUDIBLE] suggested
I would say.
Oh, don't worry about that.
We'll get to it.
We're not ready for it yet.
So in this case, I remind
somebody of something they
already know.
In this case, I tell them
they'll learn about it later.
So what do I do with mistake
number three?
That's the learning moment.
That's where I can push
the wavefront out.
Because everything's in
place to learn the
stuff at the next radius.
So if I know that the student
has made a mistake on that
wavefront, that's when I say,
this is the teaching moment.
This is when I explain
something.
So that's why it's important for
the teacher to have a good
model of where the
student is in the
initial state of knowledge.
Next thing that's important for
the teacher to know is the
way that the student learns.
Because if the student is a
computer, they can handle the
stuff in batch.
That's one thing.
If the student is a third
grader who has a limited
capacity to store stuff, then
that makes a difference in how
you teach it.
You might teach it that way to
the third grader, and that
way, buried underneath this
board, to a computer.
So you need to understand the
way that the learner--
the computational capacity
of the learner.
And there's also a need to
understand the computational
capacity of the user box down
there, because sometimes you
can be taught stuff that
you can't actually use.
So by now, most of you have
attempted to read that
sentence up there, right?
And it seems screwy, right?
It seems unintelligible,
perhaps?
It's a garden path sentence.
It makes perfectly good English,
but the way you
generally read it, it doesn't,
because you have a limited
buffer in your language
processor.
What does this mean?
You're expecting this to
be "to." Question.
But it's actually a command.
Here's the deal.
Somebody's got to give the
students their grades.
Well, we can have their
parents do it.
Have the grades given
to their students by
their parents, then.
So it's a command.
And you garden path on it,
because you have limited
buffer space in your
language processor.
So with parentheses you
can understand it.
You can learn about it.
You can see that it's good
English, but you can't
generally process that kind of
sentence without going back
and starting over.
And what about going
the other way?
Are there covenants that we have
to have here that involve
the student understanding some
things about the teacher?
Well, first thing there
is is trust.
The student has to presume that
the teacher is teaching
the student correct
information,
not lying to student.
Ratified that you're all here
because presumably you all
think that I'm not trying to
screw you by telling you stuff
that's a lie.
There's also this sort
of thing down here.
Understanding of the
teacher's style.
So you might say, well,
professor x, all he does is
read slides to us in
class, so why go?
You wouldn't be entirely
misadvised.
That's an understanding
of one kind of style.
Or you can say, well, old
Winston, he tries to tell us
something definite and convey a
family of powerful ideas in
every class.
So maybe it's worth dragging
yourself out of bed at 10
o'clock in the morning.
Those are style issues, and
those are things that the
student uses to determine how
to match the student's style
against that of the
instructor.
So that helps us to interpret or
think about differences in
style so that we can appreciate
whether we ought to
be learning that way, where
that way is the way that's
underneath down here, the way
you would teach a computer,
the way [? Malkowski ?]
taught a computer about
soybean diseases.
We can do it that way, or we
can do it this way with a
teacher who deliberately
organizes and shapes the
learning sequence for the
benefit of a student who has a
limited processing capability.
Now you're humans, right?
So think about what the machine
has to do here.
The machine--
in order to learn anything
definite in each of those
steps, the machine has to
build a description.
So it has to describe the
examples to itself.
That's unquestioned, right?
Because what it's doing is
looking at the differences.
So it can't look at the
differences unless it's got
descriptions of things.
So if you're like the machine,
then you can't learn anything
unless you build descriptions.
Unless you talk to yourself.
And if you talk to yourself,
you're building the kind of
descriptions that make
it possible for
you to do the learning.
And you say to me, I'm
an MIT student.
I want to see the numbers.
So let me show you
the numbers.
And when I'm going
to show numbers--
the numbers that I'm going to
show you show you the virtues
of talking to yourself.
So here's the experiment.
The experiment was done by a
friend of mine, Michelene Chi.
Always seems to go by
the name Mickey Chi.
There he is.
So here's the deal.
The students that she worked
with were expected to learn
about elementary physics.
801 type stuff.
And she took eight subjects,
and she had them--
she took them through a bunch of
examples and then she gave
them an examination.
So eight subjects, and so they
divide into two groups.
The bottom half and
the top half.
The ones who did better than
average and the ones who did
worse than average.
So then you can say, well,
OK, what did that mean?
You can say, how much did
they talk to themselves?
Well, that was measured by
having them talk out loud as
they solved the problems
on an examination.
So we could ask how much self
explanation was done by the
smart ones versus the
less smart ones?
And here are the results.
The worst ones-- the worst four
said about 10 things to
themselves.
The best four said about 35
things to themselves.
That's a pretty dramatic
difference.
Here's the data in a more
straightforward form.
This, by the way, points out
that the smart ones scored
twice as high as the
less smart ones.
And when we look at the number
of explanations they gave
themselves in two categories,
smart ones said three times as
much stuff to themselves
as the less smart ones.
So, as you can see, the
explanations break down into
two groups.
Some have to do with monitoring
and not with
physics at all.
They're things like,
oh hell, I'm stuck.
Or, I don't know what to do.
And the others have to
do with physics.
Things like, well, maybe I
should draw a force diagram.
Or let me write down f equals
ma, or something like that, as
physics knowledge.
I think it's interesting that
this average score is
different by a factor of two,
and the average talking to
oneself differed by
a factor of three.
Now this isn't quite there,
because what's not clear is if
you encourage somebody to talk
to themself, and they talk to
themselves more than they would
have ordinarily, does
that make them score better?
All we know is that the ones who
talk to themselves more do
score better.
But anecdotally, talking to
some veterans of 6.034,
they've started talking to
themselves more when they
solve problems, and they think
that it makes them smarter.
Now I would caution you not to
do this too much in public.
Because people can get the
wrong idea if you talk to
yourself too much.
But it does seem--
it does, in fact,
seem to help.
Now what I did last time
is I told you how
to be a good scientist.
What I'm telling you now is how
to make yourself smarter.
And I want to conclude this hour
by telling you about how
you can package your ideas so
that they have greater impact.
So I guess I could have said,
how to make yourself more
famous, but I've limited
myself to saying how to
package your ideas better.
And the reason you want to
package your ideas better is
because if you package your
ideas better than the next
slug, then you're going to get
the faculty position and
they're not.
If you say to me, I'm
going to be an
entrepreneur, same thing.
You're going to get the venture
capitalist money and
the next slug won't if you
package your ideas better.
So this little piece of work
on the arch business got a
whole lot more famous than
I ever expected.
I did it when I was young and
stupid, and didn't have any
idea what qualities might emerge
from a piece of work
that would make it well known.
I only figured it
out much later.
But in retrospect, it has five
qualities that you can think
about when you're deciding
whether your packaging of your
idea is in a form that will
lead to that idea becoming
well known.
And since there are five of
them, it's convenient to put
them all on the points
of a star like so.
So quality number one.
I've made these all into s-words
just to make them
easier to remember.
Quality number one is that
there's some kind of symbol
associated with a work.
Some kind of visual handle
that people will use to
remember your idea.
So what's the visual
symbol here?
Well, that's astonishingly easy
to figure out, right?
That's the arch.
For years without my
intending it, this
was called arch learning.
So you need a symbol.
Then you also need a slogan.
That's a kind of
verbal handle.
It doesn't explain the idea,
but it's enough of a handle
to, as Minsky would say, put you
back in the mental state
you were in when you understood
the idea in the
first place.
So what is the slogan
for this work?
Anybody have any ideas?
Pretty obvious.
What's essential to this
process working?
The ability to present an
example is very similar
[INAUDIBLE], that constitutes
a model but
isn't one of those.
STUDENT: [INAUDIBLE].
PROFESSOR PATRICK WINSTON:
So it's a near miss.
The next thing you need if your
work is going to become
well known is a surprise.
What's the surprise
with this stuff?
Well, the surprise--
everything that had been done
in artificial intelligence
having to do with learning
before this time was
precursors to neural nets.
Thousands of examples
to learn anything.
So the big surprise was that it
was possible for a machine
to learn something definite
from each of the examples.
So that now goes by the name
of one shot learning.
That was the surprise, that a
computer could learn something
definite from a single
example.
So let's see.
We've almost completed
our star.
But there are more
points on it.
So this point is the salient.
What's a salient--
what's a salient idea?
Jose, do you know what
a salient idea is?
He's too shy to tell me.
What's a salient idea?
Ah, who said important?
Wrong answer, but very good.
You're not shy.
So what does it really mean?
Yes.
STUDENT: Relative to
what somebody's
already thinking about?
PROFESSOR PATRICK WINSTON:
Relative to what somebody's
thinking about.
Not quite.
If you have a--
if you're an expert in--
yes?
STUDENT: [INAUDIBLE].
PROFESSOR PATRICK WINSTON:
Really close.
We're getting closer.
[INAUDIBLE].
Yes?
STUDENT: Maybe an idea that
wasn't obviously apparent, but
becomes apparent gradually as
somebody starts to understand?
PROFESSOR PATRICK WINSTON: We're
zeroing-- we're circling
the wagons here and
zeroing in on it.
Yes?
STUDENT: If I'm preempting what
you're about to say, it
has sort of a doorway of how you
can understand the idea.
PROFESSOR PATRICK WINSTON:
It's what?
Sorry.
STUDENT: It's sort of like
a doorway of how you
can grasp the idea.
PROFESSOR PATRICK WINSTON:
That's sort if it, too, but if
you study military history,
what's the salient on a fort?
Well, this is a good word to
have in your vocabulary
because it sort of means all of
those things, but what it
really means is something
that sticks out.
So on a fort, if this were a
fort, these would all be
salients because
they stick out.
So the salient idea is
usually important
because it sticks out.
But it's not-- the meaning is
not "important," the meaning
is "stick out."
So a piece of work becomes
more famous if it has
something that sticks out.
It's interesting.
There are theses that have been
written at MIT that have
too many good ideas.
And how can have too
many good ideas?
Well, you can have too many
good ideas if no one idea
rises above and becomes the idea
that people think about
when they think about you.
We have people on the faculty
who would have been more
famous if their theses
had fewer ideas.
It's amazing.
So this piece of work
did have a salient.
And the salient idea was that
you could get one shot
learning via the use
of near misses.
That was the salient idea.
The fifth thing, ah.
Talk more about this
in my "How to
Speak" lecture in January.
The fifth thing I like people
to try to incorporate into
their presentations
is a story.
Because we humans somehow
love stories.
We love people to
tell us stories.
We love things to be packaged
in stories.
And believe me, I think all of
education is essentially about
storytelling and story
understanding.
So if you want your idea to
be sold to the venture
capitalist, if you want to get
the faculty job, if you want
to get your book sold to a
publisher, if you want to sell
something to a customer, ask
yourself if your presentation
has these qualities in it.
And if it has all of those
things, it's a lot more likely
to be effective than
it doesn't.
And you'll end up
being famous.
Now you say to me, well, being
famous-- that sounds like the
Sloan School type of concept.
Isn't it immoral to
want to be famous?
Maybe that's a decision
you can make.
But whenever I think about the
question, I somehow think of
the idea that your ideas
are like your children.
You want to be sure that they
have the best life possible.
So if they're not packaged
well, they won't.
I'm also reminded of an evening
I spent at a soiree
with Julia Child.
Julia, and there's me.
And I have no idea how
come I got to sit
next to Julia Child.
I think they thought I was
one of the rich Winstons.
The Winston flowers, or the
Harry Winston diamonds or
something like that.
There I was, sitting next
to Julia Child.
And the interesting thing--
by the way, did you notice
I'm now telling a story?
The interesting thing about this
experience was that there
was a constant flow
of people--
happened to be all women--
people going past Ms. Child
saying how wonderful she was
to have made such an enormous
change in their life.
Must have been 10 of them.
It was amazing.
Just steady flow.
So eventually I leaned over to
her and I said, Ms. Child, is
it fun to be famous?
And she thought about it
a second and said,
you get used to it.
And that had a profound effect
on me, because you always say,
well, what's the
opposite like?
Is it fun to be ignored?
And the answer is, no, it's not
much fun to be ignored.
So yeah, it's something you can
get used to, but you can
never get used to having your
stuff ignored, especially if
it's good stuff.
So that's why I commend
to you this business
about packaging ideas.
And now you see that 6034
is not just about AI.
It's about how to
do good science.
It's how to make yourself
smarter, and how to make
yourself more famous.