FFN performs efficient test-time training on multi-hour videos by forgetting the exiting frame, anticipating the next, and adapting only when a surprise metric exceeds a dynamic threshold.
Noprop: Training neural net- works without full back-propagation or full forward-propagation.arXiv preprint arXiv:2503.24322, 2025
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
representative citing papers
IQFMs iteratively constructs deep quantum feature maps from shallow circuits via classical augmentation weights and contrastive layer-wise training, outperforming QCNNs on noisy quantum data and matching classical neural networks on image classification without variational parameter optimization.
SMT trains nonlinear RNNs by imitating one-step memory-transition labels generated by a Transformer, replacing BPTT's unrolled credit assignment with time-parallel supervised learning.
citing papers explorer
-
Forget, Anticipate and Adapt: Test Time Training for Long Videos
FFN performs efficient test-time training on multi-hour videos by forgetting the exiting frame, anticipating the next, and adapting only when a surprise metric exceeds a dynamic threshold.
-
Iterative Quantum Feature Maps
IQFMs iteratively constructs deep quantum feature maps from shallow circuits via classical augmentation weights and contrastive layer-wise training, outperforming QCNNs on noisy quantum data and matching classical neural networks on image classification without variational parameter optimization.
-
Pretraining Recurrent Networks without Recurrence
SMT trains nonlinear RNNs by imitating one-step memory-transition labels generated by a Transformer, replacing BPTT's unrolled credit assignment with time-parallel supervised learning.