New bounds for nearly-linear networks without training.
problem Generalization of neural networks close to linearity.
method Perturbation of linear networks to derive bounds.
result First non-vacuous bounds for neural nets.
On-device research index
arXiv research
A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.
168,657 papers · 148 categories
Trend · papers per month
48 results for “non-vacuous bounds”
New bounds show large language models can generalize beyond training data.
problem Generalization of large language models beyond training data.
method Compression bound derivation using prediction smoothing and SubLoRA.
result Large language models can discover generalizable regularities.
This paper tackles non-vacuous generalization bounds in ReLU networks by resolving rescaling invariances.
problem Non-vacuous generalization guarantees for ReLU networks with rescaling invariances.
method Proposes a lifted representation to resolve rescaling invariances and studies KL-based rescaling-invariant PAC-Bayes bounds.
result KL-based rescaling-invariant PAC-Bayes bounds provide tighter guarantees and resolve discrepancies in network complexity.
New bounds for shallow neural networks with deterministic parameters.
problem Developing generalisation bounds for shallow neural networks.
method PAC-Bayesian theory applied to shallow neural networks with deterministic parameters.
result Empirical non-vacuous bounds for shallow neural networks trained with vanilla SGD.
New bounds for large language models using token properties.
problem Vacuous generalization bounds for large language models.
method Martingale properties and Monarch matrices.
result Non-vacuous generalization bounds for LLMs up to 70B parameters.
Explaining how overparametrized neural networks simultaneously achieve low risk and zero empirical risk on benchmark datasets is an open problem. PAC-Bayes bounds optimized using variational inference (VI) have been recently proposed as a promising direction in obtaining non-vacuous bounds. We show empirically that thi…
This work compresses heavy-tailed weight matrices for tighter generalization bounds.
problem Empirical evidence linking heavy-tailed weight matrices to test set accuracy but lack of formal relationship with generalization bounds.
method Utilized the compression framework to show that heavy-tailed matrices can be compressed, resulting in sparse weight matrices.
result Demonstrated a non-vacuous generalization bound for compressed networks with heavy-tailed weight matrices.
Modern neural networks are highly overparameterized, with capacity to substantially overfit to training data. Nevertheless, these networks often generalize well in practice. It has also been observed that trained networks can often be "compressed" to much smaller representations. The purpose of this paper is to connect…
New bounds ensure reliable deep learning performance without model changes.
problem Certifying deep neural networks' reliability without altering them.
method Data-dependent generalization bounds that apply directly to trained models.
result Achieves meaningful generalization guarantees for large, unaltered deep networks.
From Low Intrinsic Dimensionality to Non-Vacuous Generalization Bounds in Deep Multi-Task Learningcs.LG
Deep multi-task learning benefits from low intrinsic dimensionality, leading to better generalization.
problem Improving generalization in deep multi-task learning with high-dimensional models.
method Parametrizing multi-task networks in a low-dimensional space using random expansions and weight compression.
result First non-vacuous generalization bounds for deep multi-task networks are derived.
The study improves PAC-Bayesian bounds for adversarial generative models.
problem Improving generalization bounds for adversarial generative models.
method Extending PAC-Bayesian theory to generative models, developing bounds for Wasserstein and total variation distances.
result New training objectives for Wasserstein and Energy-Based GANs.
New analysis improves generalization bounds for meta-learning.
problem Improving generalization in meta-learning algorithms.
method Information-theoretic analysis of MAML and its stochastic variant.
result Data-dependent generalization bound is tighter and non-vacuous.
New reinforcement learning bound improves generalization for sequential data.
problem Challenges in obtaining generalization guarantees for reinforcement learning due to sequential data.
method PAC-Bayesian reinforcement learning with consideration of Markov dependencies and mixing time.
result Demonstrated practical utility through PB-SAC, providing meaningful confidence certificates.
Contrastive unsupervised representation learning (CURL) is the state-of-the-art technique to learn representations (as a set of features) from unlabelled data. While CURL has collected several empirical successes recently, theoretical understanding of its performance was still missing. In a recent work, Arora et al. (2…
New algorithms minimize PAC-Bayesian C-Bound for majority voting, leading to scalable and accurate predictors.
problem Improving majority vote classifiers using PAC-Bayesian bounds.
method Directly optimizing PAC-Bayesian guarantees on the C-Bound with gradient descent.
result Self-bounding majority vote learning algorithms with scalable and accurate predictors.
We explore the family of methods "PAC-Bayes with Backprop" (PBB) to train probabilistic neural networks by minimizing PAC-Bayes bounds. We present two training objectives, one derived from a previously known PAC-Bayes bound, and a second one derived from a novel PAC-Bayes bound. Both training objectives are evaluated o…
SGD-trained deep nets have bounds on their generalization error.
problem Bounding generalization error for deep neural networks trained by SGD.
method Combining dynamical control of parameter norms and Rademacher complexity estimates.
result Explicit bounds depend on loss trajectory, work for various architectures.
The paper improves deep learning generalization bounds using PAC-Bayes compression.
problem Improving generalization bounds for deep neural networks.
method Quantizing neural network parameters in a linear subspace to develop tight generalization bounds.
result Large models can be compressed significantly, explaining Occam's razor.
The paper studies a stochastic majority vote approach to improve classifier accuracy.
problem Improving classifier accuracy over ensembles of classifiers.
method Minimizing a PAC-Bayes generalization bound with Dirichlet distributions.
result Achieves state-of-the-art accuracy and tight generalization bounds.
Improved generalization bounds for CNNs using Rademacher complexity.
problem Establishing non-vacuous generalization bounds for deep learning models.
method Rademacher complexity framework with novel contraction lemmas for high-dimensional mappings.
result Enhanced generalization bounds for a broader class of activation functions.
Optimizes SGLD noise structure for better generalization bounds.
problem Improving generalization bounds for large models trained with SGLD.
method Manipulates the noise structure in SGLD to optimize information-theoretical bounds.
result Optimal noise covariance is the square root of the expected gradient covariance under certain constraints.
New methods tackle robust reinforcement learning in sparse, corrupted data.
problem Tackles robust reinforcement learning in sparse, corrupted data.
method Proposes actor-critic methods with sparse robust estimator oracles.
result First non-vacuous guarantees in high-dimensional sparse MDPs with single-policy concentrability coverage.
Novel bounds for deep MDA algorithms improve performance and efficiency.
problem Improving performance of MDA algorithms with few target labels and pseudo labels.
method Information-theoretic tools and novel deep MDA algorithm.
result Algorithm-dependent generalization bounds for MDA.
This work tightens generalization error bounds using Wasserstein distance.
problem Improving expected generalization error bounds in machine learning.
method Introduces bounds based on Wasserstein distance for various settings.
result New, tighter bounds based on relative entropy and other information measures.
Survey on new data-dependent bounds for neural networks.
problem Generalization of overparameterized neural networks.
method Extending PAC-Bayesian theory, refining complexity terms, and replacing information-theoretic terms with stability assumptions.
result Unified template inequality and comparison of bounds.
Some theoretical improvements on the tightness of PAC-Bayes risk certificates for neural networkscs.LG
The paper improves risk certificate tightness for neural networks using PAC-Bayes bounds.
problem Improving the usability of risk certificates for neural networks based on PAC-Bayes bounds.
method Theoretical contributions including KL divergence bounds, efficient methodology for optimization, and methods for optimizing non-differentiable objectives.
result First non-vacuous generalization bounds on CIFAR-10 for neural networks.
Paper improves generalization bounds for noisy stochastic algorithms.
problem Improving generalization bounds for noisy stochastic algorithms.
method Introduces Exponential Family Langevin Dynamics (EFLD) and establishes data-dependent expected stability based generalization bounds.
result Sharp generalization bounds with O(1/n) sample dependence and gradient discrepancy.
Proposes new priors for neural networks to improve generalization and uncertainty.
problem Improving generalization and uncertainty estimation in neural networks.
method Exploits scalable and structured posteriors as priors with generalization guarantees.
result Improves generalization and uncertainty estimation with non-vacuous bounds.
Survey and compare PAC-Bayes bounds for bandit problems.
problem Designing and evaluating bandit algorithms with strong performance guarantees.
method PAC-Bayes bounds applied to bandit problems.
result PAC-Bayes bounds useful for offline bandit algorithms, but loose for online algorithms.
This paper provides theoretical insights into why and how deep learning can generalize well, despite its large capacity, complexity, possible algorithmic instability, nonrobustness, and sharp minima, responding to an open question in the literature. We also discuss approaches to provide non-vacuous generalization guara…
New bounds show robust models can generalize well, contrary to prior theories.
problem Existing robustness-based error bounds are vacuous for the best classifier.
method Developed novel bounds that converge to the true error of the best classifier.
result New bounds converge to the true error of the best classifier, improving generalization.
Machine learning systems must adapt to data distributions that evolve over time, in applications ranging from sensor networks and self-driving car perception modules to brain-machine interfaces. We consider gradual domain adaptation, where the goal is to adapt an initial classifier trained on a source domain given only…
New method gives provable error bounds for neural nets under distribution shift.
problem Proving reliable error bounds for neural networks under distribution shift.
method Optimizing a classifier to disagree with another, using a new 'disagreement loss'.
result Valid error bounds with comparable accuracy to competitive methods.
Study shows gap between uniform convergence and test error in random feature models.
problem Understanding the gap between uniform convergence and test error in random feature models.
method Analytical expressions for uniform convergence over norm balls, interpolators, and minimum norm interpolator risk derived and proved.
result Uniform convergence over interpolators still gives a non-trivial bound of test error even when classical uniform convergence is vacuous.
New bounds predict deep learning generalization better than existing methods.
problem Predicting generalization errors for deep learning models.
method Function-based PAC-Bayesian bounds that meet multiple desiderata.
result The new bound performs significantly better than existing parameter-based PAC-Bayes bounds.
New bounds for transfer learning in linear models, improving generalization.
problem Understanding when auxiliary data helps in improving generalization in linear models.
method Derivation of exact error bounds and optimal task weights for linear regression and linear neural networks.
result First non-vacuous sufficient conditions for beneficial auxiliary learning in linear neural networks.
This work analyzes generalization in federated learning using information theory.
problem Generalization performance in federated learning is less explored compared to centralized learning.
method The work applies an information-theoretic analysis via the conditional mutual information (CMI) framework to study federated learning's two-level generalization.
result The work derives multiple CMI-based bounds, including hypothesis-based CMI bounds and fast-rate evaluated CMI bounds, which improve convergence rates for specific model aggregation strategies and structured loss functions.
TRM improves long-horizon LLM RL by masking divergent sequences.
problem Long-horizon reinforcement learning with LLMs suffers from off-policy mismatch and approximation errors.
method Derives and applies trust region bounds to control divergence, proposing Trust Region Masking.
result First non-vacuous monotonic improvement guarantees for long-horizon LLM-RL.
TRM improves long-horizon reinforcement learning for LLMs by masking divergent sequences.
problem Long-horizon reinforcement learning for LLMs suffers from off-policy mismatch and approximation errors.
method Derives and applies trust region bounds to control divergence, proposing Trust Region Masking.
result First non-vacuous monotonic improvement guarantees for long-horizon LLM-RL.
New bounds improve neural network generalization through slicing.
problem Difficulty in evaluating mutual information in high dimensions for neural networks.
method Slicing the parameter space and using disintegrated mutual information and k-sliced mutual information.
result Slicing improves generalization and offers significant computational and statistical advantages.
New method uses unlabeled data to improve generalization bounds for deep learning.
problem Vacuous guarantees and shrinking holdout sets for overparameterized models.
method Augmenting labeled training set with unlabeled data and training as usual.
result Proves tight upper bounds on true risk for 0-1 empirical risk minimization.
Paper develops tighter risk certificates for contrastive learning models.
problem Statistical theory for contrastive learning is lacking, especially for practical models like SimCLR.
method Develops non-vacuous PAC-Bayesian risk certificates considering practical SimCLR factors.
result Risk certificates for contrastive loss and downstream prediction are much tighter than previous results.
New bound proves rationality helps generalization in self-supervised learning.
problem Proving generalization gap in self-supervised learning.
method Proving upper bound on generalization gap for classifiers using rationality and self-supervised representations.
result Generalization gap tends to zero if classifier complexity is low relative to number of samples.
Data structure affects deep learning performance, study finds.
problem Understanding why deep learning performs poorly on typical datasets.
method Analyzed input correlation matrices and network Hessians, developed PAC-Bayes bounds.
result Sloppy eigenspectra in input data correlate with poor deep learning performance.
Develops PAC-Bayesian framework for physics-informed machine learning.
problem Lack of statistical generalisation understanding for PIML models.
method PAC-Bayesian framework with multi-task perspective, incorporating physical structure.
result High-probability generalisation guarantees with unbounded losses.
Paper presents new training methods for neural networks with tighter risk certificates.
problem Training probabilistic neural networks with tighter risk certificates.
method Derived from PAC-Bayes bounds, two training objectives implemented for the first time in neural networks.
result Competitive test set errors and non-vacuous risk bounds with tighter values than previous results.
Unified understanding of neural networks on group operations verified.
problem Understanding and verifying neural networks trained on group operations.
method Investigated one-hidden-layer neural networks trained on binary operation of finite groups, revealing structure and providing a compact proof of model performance.
result Verified explanation applies to a large fraction of networks trained on the symmetric group S5, providing a >=95% accuracy bound for 45% of models.
The paper proves D-SGD's stability and generalization bound, highlighting the importance of communication topology.
problem The stability and generalization of decentralized stochastic gradient descent (D-SGD).
method Theoretical analysis of D-SGD's stability and generalization bound, considering spectral gap and communication topology.
result D-SGD's generalization bound is positively correlated with the spectral gap of the communication topology.