1
00:00:00,000 --> 00:00:07,301
Welcome to the final lecture on Predict.
This week, we'll start with bottom up

2
00:00:07,301 --> 00:00:12,026
prediction.
How we can predict future values of data

3
00:00:12,026 --> 00:00:17,978
from the data that we have.
We've seen some of this already in

4
00:00:17,978 --> 00:00:22,548
learning.
How we can predict what future instances

5
00:00:22,548 --> 00:00:27,118
look like or are part of based on past
experience.

6
00:00:27,118 --> 00:00:33,881
But we'll go on to deal with predicting
values rather than simply classes.

7
00:00:33,881 --> 00:00:40,827
We'll begin with the very basic prediction
technique, which is least-squares

8
00:00:40,827 --> 00:00:45,580
approximation, and function approximation,
in general.

9
00:00:46,580 --> 00:00:53,026
Then, we'll go on to show the relationship
between prediction, optimization, and then

10
00:00:53,026 --> 00:00:57,220
controlling what you want to do with your
predictions.

11
00:00:57,220 --> 00:01:04,965
Next, we'll ask how the brain actually
does all these things in a very smooth and

12
00:01:04,965 --> 00:01:11,358
almost invisible manner and talk about
some recent developments in something

13
00:01:11,358 --> 00:01:17,778
called hierarchical temporal memory, which
is a prediction system modeled after how

14
00:01:17,778 --> 00:01:21,567
the brains and neurons actually are put
together.

15
00:01:21,567 --> 00:01:27,831
And it is the latest advance in what was
originally the field of neural networks

16
00:01:27,831 --> 00:01:34,173
but we'll see that neural networks, belief
networks and everything is sort of coming

17
00:01:34,173 --> 00:01:38,350
together in this fairly interesting
development.

18
00:01:38,350 --> 00:01:43,993
However, it turns out that bottom up
prediction so far, can go only a certain

19
00:01:43,993 --> 00:01:49,462
way and one needs to combine many
different techniques including symbolic

20
00:01:49,462 --> 00:01:53,232
reasoning as well as direct learning from
the data.

21
00:01:53,232 --> 00:01:58,554
And we will discuss a very popular and
fairly old architecture called the

22
00:01:58,554 --> 00:02:04,319
blackboard architecture which is becoming
more and more important, as all these

23
00:02:04,319 --> 00:02:08,680
techniques start working together in large
complex systems.

24
00:02:09,680 --> 00:02:15,238
Of course then, we'll finally summarize
seeing where we've come in this course

25
00:02:15,238 --> 00:02:20,298
about what we've learned about web
intelligence, the brain, and adaptive

26
00:02:20,298 --> 00:02:24,004
business intelligence based on all these
techniques.

27
00:02:24,004 --> 00:02:29,491
And lastly, I'll leave you with some
challenge problems, which are fairly deep

28
00:02:29,491 --> 00:02:34,480
and might interest some of you to actually
take these up for research.