arXiv:2512. 23043v2 Announce Type: replace Abstract: Federated Averaging (FedAvg) often degrades under non-IID client data, but it remains unclear whether this degradation reflects the loss of client-learned representations or a failure to use representations that are still present.
arXiv:2605. 20282v3 Announce Type: replace-cross Abstract: Machine unlearning in Vertical Federated Learning (VFL) has attracted growing interest, yet existing methods certify forgetting solely using output-level metrics.
arXiv:2605. 20282v2 Announce Type: replace-cross Abstract: Machine unlearning in Vertical Federated Learning (VFL) has attracted growing interest, yet existing methods certify forgetting solely using output-level metrics.
arXiv:2606. 12138v1 Announce Type: cross Abstract: Sparse autoencoders (SAEs) are widely used to interpret neural network representations, but their utility depends on whether the learned features are reproducible across training runs.
arXiv:2507. 07754v3 Announce Type: replace-cross Abstract: Machine unlearning is usually evaluated by what the classifier outputs: forget-set accuracy, confidence, membership-inference scores.
arXiv:2607. 12438v1 Announce Type: new Abstract: Deep networks trained with label noise often learn clean structure before memorizing corrupted labels.
arXiv:2609.24379v1 Announce Type: cross Abstract: Mechanistic interpretability of vision transformers seeks to decompose model computation into human-readable units, but learned representations entan...
Mechanistic interpretability of vision transformers seeks to decompose model computation into human-readable units, but learned representations entangle many concepts in each neuron. Feature superposi...
arXiv:2607.01630v2 Announce Type: replace Abstract: Dynamic expansion methods for class-incremental learning (CIL) protect task-specific knowledge by growing dedicated tokens or subnetworks, yet our...
Direct feedback alignment (DFA) trains hidden layers via fixed random projections of output error, but with tanh hidden units and independent sigmoid outputs, plain stochastic gradient descent can stall near a constant predictor of class frequencies. This stall is traced to the error’s common mode—a rank‑one component shared across inputs—that drives tanh units toward saturation. The study shows that calibration of the baseline readout to class priors suppresses collapse and speeds learning, while other interventions such as using Adam, adjusting feedback strength, or subtracting batch means affect the severity and recovery of collapse across MNIST, CIFAR‑10, and deeper networks.
arXiv:2602.01718v2 Announce Type: replace Abstract: Predicting generalization from quantities available before target-test evaluation remains a central challenge in deep learning. The systematic benc...
Dynamic expansion methods for class-incremental learning (CIL) protect task-specific knowledge by growing dedicated tokens or subnetworks, yet our analyses suggest that classification supervision alone does not sufficiently preserve task-agnostic shared backbone representations over long incremental sequences. We identify two intertwined challenges: cross-task confusion from sequential training on predominantly current-task data, which biases decision boundaries toward recent tasks; and under-optimized shared representations in the backbone that cap long-term discriminability as tasks accumulate.
arXiv:2607. 21353v1 Announce Type: new Abstract: Machine unlearning aims to remove the influence of specific training data while preserving model utility.