collapse
A non-Schrodinger process in which the wave function abruptly reduces to a definite outcome, as opposed to smooth evolution.
As defined in 241 | Tim Maudlin on Locality, Hidden Variables, and Quantum Foundations
Where it comes up · 2
- Yann LeCun: Special Lecture on AI and World Models4 claims
- 241 | Tim Maudlin on Locality, Hidden Variables, and Quantum Foundations1 claim
Most important claims · 5 of 5Search all claims for “collapse” →
- If collapse of the wave function must be triggered by 'measurement' or 'observation,' you inherit the unanswerable question of what counts as an observer — can a mouse or amoeba trigger collapse? — which pushes some, like Wigner, toward invoking consciousness and the mind-body problem.241 | Tim Maudlin on Locality, Hidden Variables, and Quantum Foundations
- Sigreg (Sketch anisotropic Gaussian regularization) prevents JEPA collapse by maximizing information content in encoder outputs through a regularization method based on projecting distributions along multiple axes and ensuring marginal distributions match isotropic Gaussians.Yann LeCun: Special Lecture on AI and World Models
- Geometric deep learning (graph neural networks, manifold learning) should be applied to JEPA: training JEPA architectures on graph-structured inputs has not been done yet but would be very interesting and promising to explore.Yann LeCun: Special Lecture on AI and World Models
- Training JEPA architectures faces the 'collapse' problem: if you simply minimize prediction error in representation space, the system can cheat by predicting constant representations, making the prediction trivial and useless.Yann LeCun: Special Lecture on AI and World Models
- The I-JEPA method prevents collapse through momentum-based weight updates—maintaining a second encoder whose weights are an exponential moving average of the primary encoder's weights, which unexpectedly prevents representation collapse, though the theoretical explanation remains incomplete.Yann LeCun: Special Lecture on AI and World Models