[LeCun] A path towards autonomous machine intelligence

Aug 24, 2022

Yann LeCun’s new position paper titled “A path toward autonomous machine learning”.

LeCun presents three main challenges:

1. How can machines learn to represent the world, learn to predict, and learn to act largely by observation?

2. How can machine reason and plan in ways that are compatible with gradient-based learning?

3. How can machines learn to represent percepts and action plans in a hierarchical manner, at multiple levels of abstraction, and multiple time scales?

Then, he proposes the following:

1. An overall cognitive architecture in which all modules are differentiable and many of them are trainable.

2. JEPA and Hierarchical JEPA: a non-generative architecture for predictive world models that learn a hierarchy of representations.

3. A non-contrastive self-supervised learning paradigm that produces representations that are simultaneously informative and predictable.

4. A way to use H-JEPA as the basis of predictive world models for hierarchical planning under uncertainty.

Choe: Some thoughts:

1. Does the brain really build world models?
2. Should everything really be graident-based?

https://openreview.net/pdf?id=BZ5a1r-kVsf

← Back to all articles         Quick Navigation:    Next:[ j ] – Prev:[ k ] – List:[ l ]