1
00:00:00,000 --> 00:00:07,532
Welcome back.
We'll now begin our lectures again with

2
00:00:07,532 --> 00:00:12,495
Learn.
We'll start by revisiting the concept of

3
00:00:12,495 --> 00:00:19,886
learning which we studied during the
listen lecture where we learned about the

4
00:00:19,886 --> 00:00:25,687
naive base classifier.
I will try to put that on a more firmer,

5
00:00:25,687 --> 00:00:31,846
formal footing as well as,
Make it generalized to other forms of

6
00:00:31,846 --> 00:00:35,962
learning.
In particular, unsupervised learning where

7
00:00:35,962 --> 00:00:40,885
we don't necessarily have a training set
like we did earlier.

8
00:00:40,885 --> 00:00:47,100
We'll learn how to learn business rules or
other kinds of rules from data.

9
00:00:47,580 --> 00:00:54,304
We'll also figure out how to learn the
classes and features themselves, and study

10
00:00:54,304 --> 00:01:00,696
how this happens in recommender systems
such as book recommendations, product

11
00:01:00,696 --> 00:01:05,760
recommendations, movie reviews, and stuff
like that on the web.

12
00:01:07,280 --> 00:01:14,392
We will also speculate on how this tells
us something about how human beings learn

13
00:01:14,392 --> 00:01:22,130
basic concepts about the world without
ever being taught explicitly what features

14
00:01:22,130 --> 00:01:28,215
to look for.
Next, we'll turn to supervised learning

15
00:01:28,215 --> 00:01:35,783
again where we'll figure out how to learn
facts, such as, antibiotics kill bacteria,

16
00:01:35,783 --> 00:01:43,166
or Einstein was born in Ulm, and other
things that we can learn formally from

17
00:01:43,166 --> 00:01:48,980
large collections of text such as that is
available on the web.

18
00:01:50,360 --> 00:01:56,184
We'll then speculate on whether knowledge
learned in this manner.

19
00:01:56,184 --> 00:02:01,650
Whether it's by, by crawling through large
volumes of text.

20
00:02:01,650 --> 00:02:06,220
Or learning classes an features from data
or rules.

21
00:02:06,860 --> 00:02:13,659
Actually, constitutes knowledge of any
kind and, philosophically, does it give us

22
00:02:13,659 --> 00:02:20,320
some satisfaction that we've figured out
something about how we all learn so much

23
00:02:20,320 --> 00:02:25,825
about the world.
So, gets, get ready.

24
00:02:26,216 --> 00:02:36,380
We'll be delving deeper and covering a
large tract of material this week.

25
00:02:38,160 --> 00:02:44,144
We'll also be doing things a bit more
formally, a little more notation, some

26
00:02:44,144 --> 00:02:50,202
mathematical notation.
But, for those of you who get confused,

27
00:02:50,202 --> 00:02:58,544
hopefully you won't. But, please remember
that in the end, it's all about counting.

28
00:02:58,544 --> 00:03:06,156
So, don't get scared of the notation.
Be careful to remember what it's all

29
00:03:06,156 --> 00:03:10,848
about.
Counting, frequencies, probabilities, and

30
00:03:10,848 --> 00:03:14,540
essentially,
Arithmetic.

31
00:03:15,020 --> 00:03:18,880
So,
Let's get started.