arXiv:2602. 24266v2 Announce Type: replace-cross Abstract: Which internal mechanisms of a neural network can be replaced while preserving the computation it performs?
arXiv:2609.18239v1 Announce Type: new Abstract: Structured pruning is commonly formulated as ranking individual channels, although channel responses can be complementary or cancel through downstream...
arXiv:2607. 27255v1 Announce Type: cross Abstract: Neural networks increasingly combine data across populations, time periods, and operating conditions to improve generalization.
Matryoshka Attribution (MAttr) is a mask‑learning method that identifies nested subsets of a language model’s internal components by minimizing downstream loss. It uses a differentiable sigmoid top‑k operator and randomizes sparsity during training to produce an attribution ordering of components. MAttr tops the Mechanistic Interpretability Benchmark leaderboard and can be applied via reinforcement learning to pinpoint weight changes that control behaviors such as refusal in Llama 3.1 8B Instruct, where restoring just 1% of weights removes refusals while preserving capabilities.
Neural networks increasingly combine data across populations, time periods, and operating conditions to improve generalization. This raises a reliability question: whether a model refitted on pooled data preserves an action ordering supported by both sources.
arXiv:2607. 20201v1 Announce Type: cross Abstract: Additive models buy interpretability by forbidding feature interactions, a constraint that neural instantiations enforce architecturally.
arXiv:2603. 01372v2 Announce Type: replace-cross Abstract: Concept Bottleneck Models (CBMs) enhance the interpretability of end-to-end neural networks by introducing a layer of concepts and predicting the class label from the concept predictions.
arXiv:2605.06240v2 Announce Type: replace-cross Abstract: Forward-Forward (FF) training lets each layer learn from a local goodness criterion. In cumulative-goodness variants, later layers can inheri...
CrossGMN introduces a graph metanetwork that processes a trained source network and an initialized target network simultaneously, enabling equivariant cross‑architecture weight‑space transformations. By preserving symmetry through cross‑network message passing, CrossGMN can refine target network initializations while remaining invariant to source permutations and equivariant to target permutations. Experiments demonstrate that CrossGMN accelerates knowledge distillation, transfers across datasets without retraining, and unifies compression from diverse source architectures into a common target architecture.
The paper discusses how backpropagation enables deep learning but does not inherently organize parameters for reusable functional components, leading to weight entanglement where overlapping parameter sets hinder independent modification. It introduces weight operators—parameterized modules that can be composed at inference—to address this, proposing a two-stage learning process that first infers operator composition and then updates only the selected operators. Vector Networks (VNs) are presented as an implementation that couples operator selection to local error-driven updates, demonstrating that learned operators can be recombined in unseen ways while keeping updates confined to the relevant parameter sets.
The paper introduces a Hybrid Quadratic Unconstrained Binary Optimization (QUBO) framework for structured neural network pruning that integrates task‑aware sensitivity metrics (first‑order Taylor and Weight‑Fisher) into the objective’s linear term and optionally uses activation similarity for quadratic interactions. It controls pruning cardinality via a binary search over a capacity incentive rather than an explicit penalty and further refines the pruning mask with a two‑stage QUBO–Tensor‑Train strategy that employs gradient‑free black‑box optimization. Experiments on SIDD image denoising with a Half‑UNet model demonstrate that this Hybrid QUBO outperforms Taylor and L1‑based QUBO baselines in PSNR and SSIM, while also revealing computational and deployment challenges of mask‑based pruning.
arXiv:2607. 09967v1 Announce Type: cross Abstract: Many neural networks operations have a multiplicative nature rather than additive: halving or doubling a norm are analogous relatively but require unequal optimization distances when taking linear steps.
arXiv:2606. 07615v1 Announce Type: cross Abstract: Deep neural networks often contain redundant hidden units.