1. "Building Machines That Learn
and Think Like People", Lake (BM)
2. By simple syntactic counting, five
conceptual fields emerge in the article:
a. Learning (420), learning-to-learn
(26), cognit (218)
b. Model: generative (41), generaliz (34),
model (265), theor (51),
c. Transfer (18), reuse (4), new (137),
novel (26), re- (3)
d.
Concept: compositional (45), structur (59),
abstrac (11), inductive (28), concept (92)
e. Few data (23), fast (16), one-shot
(12)
3. Which reads: learning-to-learn is the human learning / cognitive mode,
which builds cross-domain transferable models / theories, precisely thanks to
abstract / conceptual approach, requiring little data to learn a new domain
4. BM essentially gives two examples,
which we identify with 2 categories:
a. Lego
(characters challenge): parts or components and their concatenation rules (CF
cat Monoid in Spivak’s “Category Theory for the Sciences”)
b. Agnt
(Frostbite): agents, endowed with objectives, intentional and rational actors
5. BM opposes:
a. Deep Learning : learning by heart,
agnostic, data-intensive and computational, non-transferable
b. Human: conceptual, theoretical /
modeling, reusable in other fields
6. the DQN learning Frostbite
undoubtedly has high-level features equivalent to agents (hostile or not) but
it does not have the general conceptual grasp of what is an agent, which would
allow it to move easily from one game to another (transfer), unlike a human
7. What the human does is exactly to
construct (more or less easily) a functor of a new (for him) domain towards a more
or less abstract category; this is what happens in the functor
Characters challenge → Lego
This structural point of view, the one defended
in the Erlangen program, puts relations between objects above the objects
themselves, whereas Deep Learning does not explicitly distinguish objects and
relations.
8. As already said, 7 is not automatic,
and this search goes through heuristics
9. Solving problems' heuristics:
special case of maths:
a. Polya "how to solve it"
b. Terence Tao "solving
mathematical problems: a personal perspective": 62 occurrences of the word
"strategy" (in fact Tao prepared for the Olympics by reading Polya)
c. "Learning mathematics using
heuristic approach", Hoon
d. "Methodix" collection
e. …
10. Let us attach an euPEDIa category,
and see it as essentially isomorphic to Grph.
For each student, facing a pb bp, there is thus an optimization of the
sequence of heuristics {h(t)} adapted to his personality :
Learn: (student, pb) → {h(t)}
11. Although the idea of "school
according to Watson (India)" makes its way
It is not obvious to find an automatic learning as in 10
No comments:
Post a Comment