NLP Highlights

116 - Grounded Language Understanding, with Yonatan Bisk

Jul 3, 2020
Ask episode
Chapters
Transcript
Episode notes
1
Introduction
00:00 • 3min
2
The Space of Tasks Is Quite Wide
03:19 • 3min
3
The Miss Maton Semantics
05:54 • 5min
4
How Do We Learn These Abstractions?
10:25 • 3min
5
Is There a Future for Machine Learning?
12:56 • 2min
6
Is Self Play Enabled Learning a Grounding Problem?
14:59 • 2min
7
Alfred
16:58 • 4min
8
The Alfre Data Set Is Different From Prior Grounding Data Sets
21:00 • 3min
9
Aa Navigation Versus Gratis Base Navigation?
23:37 • 3min
10
Can You Give an Example of a Problem, an Anted?
27:02 • 2min
11
The Simplest Way to Get the Shortest Path Trajectory in a Second Bot?
29:12 • 5min
12
How to Draw a Model for a Sink Faucet?
33:58 • 2min
13
How Do You Sort of Tate Plans From Abstract Goals?
35:54 • 2min
14
Therst, I'd Like the Agent to Just Get the High Level Actions
37:32 • 3min
15
How to Make a Model for a Competition?
40:49 • 2min
16
How to Predict the Next Action in a Language Instruction
42:22 • 4min
17
The Average Number of Steps in Machine Translation
46:07 • 3min
18
Do the High Level Goalds Perform as Well as Those With Cester Level Gods?
48:37 • 4min
19
What Does It Mean to Build Intermediary Representations?
53:02 • 4min
20
Is There Anything You Want to Talk About?
56:44 • 3min