video thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Bayesian models of human inductive learning

Published on Feb 4, 202527433 Views

In everyday learning and reasoning, people routinely draw successful generalizations from very limited evidence. Even young children can infer the meanings of words, hidden properties of objects, or t

Related categories

Presentation

Bayesian models of human inductive learning Josh Tenenbaum MIT Department of Brain and Cognitive Sciences Computer Science and AI Lab (CSAIL)01:40
Lab members46:40:00
The probabilistic revolution in AI 55:33:20
Learning concepts from examples110:00:00
Everyday inductive leaps129:10:00
The solution01168:53:20
The approach: from statistics to intelligence236:06:40
The “shape bias” in word learning (Landau, Smith, Jones 1988)286:40:00
Is the shape bias learned?315:00:00
Transfer to real-world vocabulary350:16:40
Learning about feature variability401:40:00
Learning about feature variability01414:26:40
A hierarchical Bayesian model423:36:40
A hierarchical Bayesian model01435:16:40
A hierarchical Bayesian model02453:36:40
A hierarchical Bayesian model03464:26:40
A hierarchical Bayesian model04480:33:20
Learning the shape bias493:03:20
Learning the shape bias01506:23:20
Learning to transfer selectively550:33:20
Learning to transfer selectively01570:00:00
Property induction608:03:20
The computational problem633:03:20
P(D|S): How the structure constrains the data of experience706:06:40
P(D|S): How the structure constrains the data of experience01714:43:20
P(D|S): How the structure constrains the data of experience02734:10:00
slide34749:10:00
[c.f., Lawrence, 2004; Smola & Kondor 2003]752:46:40
Structure S761:06:40
Cows have property P. Elephants have property P. Horses have property P.769:10:00
Testing different priors791:23:20
Learning about spatial properties 821:40:00
Discovering structural forms845:33:20
Discovering structural forms01851:06:40
People can discover structural forms856:40:00
The ultimate goal881:06:40
A “universal grammar” for structural forms888:53:20
Hierarchical Bayesian Framework912:46:40
slide47925:33:20
Structural forms from relational data956:23:20
Lab studies of learning structural forms958:36:40
Development of structural forms as more data are observed960:00:00
Beyond “Nativism” versus “Empiricism”1002:30:00