Statistical field theory aids in understanding deep learning complexities.
problem Complexity and lack of theoretical understanding in deep learning.
method Statistical field theory as a theoretical framework.
result Field theory provides insights into generalization, bias, and feature learning.
Theory explains how deep nets learn features from data.
problem Understanding how deep neural networks learn features from data.
method Developed a noise-nonlinearity phase diagram and a mechanical theory.
result Links feature learning across layers to generalization.
Unified theory of deep learning from approximation to emergence.
problem Understanding the mechanisms behind deep learning.
method Unified, proof-oriented approach tracing from classical foundations to contemporary mechanisms.
result Unified theory explaining deep learning from approximation to emergence.
Deep learning networks are approximated using dynamical systems theory.
problem Understanding the approximation capabilities of deep learning networks.
method Modeling deep residual networks as continuous-time dynamical systems and using approximation theories in Lp. result Established general sufficient conditions for universal approximation of deep residual networks.
Deep ReLU networks can approximate matrix-vector products with error bounds.
problem Can deep ReLU networks accurately approximate matrix-vector products?
method Derived error bounds in Lebesgue and Sobolev norms for deep ReLU FNNs.
result Developed deep approximation theory with successful applications.
Lecture notes on linear neural networks for deep learning optimization and generalization.
problem Understanding optimization and generalization in deep learning models.
method Mathematical tools and dynamical systems theory.
result Potential of mathematical tools to enhance understanding of deep learning.
This paper introduces a novel measure-theoretic theory for machine learning that does not require statistical assumptions. Based on this theory, a new regularization method in deep learning is derived and shown to outperform previous methods in CIFAR-10, CIFAR-100, and SVHN. Moreover, the proposed theory provides a the…
A theoretical framework for deep learning is proposed to explain its effectiveness.
problem Lack of a comprehensive theory explaining deep learning's effectiveness.
method Integrates three characteristics into a graphical model called neurashed.
result Explains common empirical patterns in deep learning and provides insights into regularization and elasticity.
Unified deep learning theory via dynamical systems and optimal control.
problem Lack of a unified framework in deep learning theory.
method Viewing deep neural networks as discrete-time nonlinear dynamical systems and optimization algorithms as controllers.
result Revealed convergence and generalization properties of training processes.
We propose using category theory to unify deep learning architectures.
problem Lack of a coherent bridge between model constraints and implementations.
method Apply category theory to unify neural network design.
result Theory recovers constraints from geometric deep learning and encodes standard constructs.
Improves deep learning theory by reducing over-parametrization size.
problem Deep learning theory over-parametrization issues.
method Used Matrix Chernoff Bound to improve over-parametrization size.
result Improved over-parametrization size over previous results.
Lectures on deep learning from a learning theory perspective.
problem Understanding how deep learning architectures lead to inductive bias.
method Statistical learning theory and stochastic optimization.
result Gradient descent on linear diagonal networks can lead to various forms of implicit bias.
Geometric deep learning predicts knot invariants.
problem Predicting knot invariants from knot data.
method Constructing a functor from knots to graphs and using graph neural networks.
result High generalization capabilities demonstrated.
Simplified neural network EFTs reveal a single critical condition.
problem Understanding neuron statistics in neural networks at initialization.
method Diagrammatic approach to effective field theories (EFTs).
result A single condition governs criticality of all neuron preactivations.
A theory of deep learning is emerging, focusing on training dynamics and statistics.
problem Develop a scientific theory to understand deep learning.
method Synthesize research into five areas: idealized settings, tractable limits, mathematical laws, hyperparameters, and universal behaviors.
result The emerging theory is a mechanics of the learning process, named learning mechanics.
Theoretical analysis improves understanding of Deep Q-Learning's behavior.
problem Lack of formal guarantees and gaps between theory and practice of Deep Q-Learning.
method Dynamical systems perspective, focusing on realistic assumptions.
result Proves convergence of Deep Q-Learning under specific conditions.
Much attention has been devoted recently to the generalization puzzle in deep learning: large, deep networks can generalize well, but existing theories bounding generalization error are exceedingly loose, and thus cannot explain this striking performance. Furthermore, a major hope is that knowledge may transfer across …
A novel model combines deep learning and extreme value theory for multivariate cyber risk prediction.
problem High dimensionality and heavy tails in multivariate cyber risk patterns.
method Combines deep learning for point predictions and extreme value theory for quantile predictions.
result The model provides satisfactory high quantile predictions and accurate point predictions.
Book introduces deep learning methods with math, theory, and applications.
problem Understanding deep learning algorithms and their mathematical foundations.
method Reviews various ANN architectures and optimization methods, covers theoretical aspects.
result Provides a solid mathematical foundation for deep learning.
This paper organizes recent deep learning theory advances.
problem Lack of theoretical foundations in deep learning.
method Literature review and categorization into six groups.
result Organized recent advances in deep learning theory.
Simplifies deep learning scaling analysis without sacrificing accuracy.
problem Interpreting feature learning mechanisms and determining network implicit bias in high-dimensional settings.
method Developed a heuristic approach for predicting data and width scales of feature learning patterns.
result Predictions align with known results and extend to complex architectures.
Deep-IRT combines deep learning and IRT for explainable knowledge tracing.
problem Lack of explainability in deep learning-based knowledge tracing models.
method Synthesis of DKVMN and IRT models to estimate student and item parameters.
result Deep-IRT retains DKVMN performance while providing psychological interpretations.
New theory predicts deep neural network learning curves.
problem Understanding and optimizing deep neural networks.
method Gaussian field theory and renormalization group.
result Accurate predictions of deep neural network learning curves.
Paper explores Monge-Ampère in deep learning and quantum geometry.
problem Understanding the Monge-Ampère equation in deep learning.
method Review of Boltzmann learning, connection to optimal transport, insights from quantum geometry, renormalization group flow.
result Space of covariance matrices in learning dynamics coincides with the CAH cone.
Deep learning models viewed through tame geometry for convergence guarantees.
problem Understanding convergence guarantees in deep learning models.
method Introducing tame geometry concepts and tools for nonsmooth nonconvex settings.
result Illustrates tame geometry as a natural framework for AI systems, especially deep learning.
Deep learning explained through spectral filtering of hierarchical features.
problem Understanding how deep neural networks learn useful representations from data.
method Neural Low-Degree Filtering (Neural LoFi) as a stylized limit of gradient-based training.
result Predicts how representations are selected layer by layer and explains emergence of concepts.
Bayesian regularizations are explicitly implemented in CNNs, improving deep learning generalization.
problem Improving generalization in deep learning models.
method Introduced a novel probabilistic representation for CNN hidden layers and demonstrated their Bayesian nature.
result CNNs have explicitly Bayesian regularizations based on Bayesian regularization theory.
This paper improves causal inference using deep neural networks for low-dimensional covariates.
problem Improving causal inference with deep learning for high-dimensional covariates.
method Doubly robust off-policy learning with deep neural networks on low-dimensional manifolds.
result Nonasymptotic regret bounds for finite- and continuous-action scenarios, converging at a fast rate depending on intrinsic manifold dimension.
Deep learning and set theory improve prediction accuracy regardless of data relevance.
problem Improving prediction accuracy with limited relevant training data.
method Deep learning and set theory applied to large labeled training data.
result Exceptional prediction results achieved with irrelevant training data.
Develops deep learning model for detecting anomalies in transportation data.
problem Anomaly detection in temporal data of transportation networks.
method Proposes EVT-LSTM model combining LSTM and EVT, trained with an objective function.
result EVT-LSTM model outperforms other models in anomaly detection.
The paper explains emergent phenomena in deep learning using entropic forces.
problem Understanding the cause of emergent phenomena in deep learning and large language models.
method Proposes a rigorous entropic-force theory for neural networks trained with SGD and variants.
result Shows that representation learning is governed by emergent entropic forces that break continuous symmetries and preserve discrete ones.
This paper explores how deep learning models can fit data exactly and why this is important.
problem Understanding why deep learning models can fit data exactly and generalize well.
method Interpolation and over-parameterization as key themes to understand deep learning.
result Interpolation and over-parameterization are crucial for deep learning models to fit data exactly and generalize well.
Deep neural networks can approximate invariant/equivariant functions with fewer parameters.
problem Approximating functions that respect group symmetries with neural networks.
method Constructing deep neural networks with G-actions and G-equivariant/invariant affine transformations. result Deep neural networks can approximate G-invariant/equivariant functions with exponentially fewer parameters. Unified theory linking Bayesian and ensemble methods in deep learning.
problem Uncertainty quantification in deep learning.
method Reformulating optimisation as convex optimisation in probability measures, studying Wasserstein gradient flows.
result Unified theory explaining success of deep ensembles over variational inference.
Theory proposes neural networks can be initialized for optimal information transmission.
problem Optimizing neural networks for optimal information transmission and representation.
method Developed a corrected mean-field framework to study neural networks as information channels, proving mutual information maximization at dynamic isometry.
result Mutual information maximization is realized between inputs and propagated signals when neural networks are initialized at dynamic isometry.
Deep learning quantifies butterfly phenotypes, validating evolutionary theory.
problem Capturing comprehensive phenotypic information of butterflies.
method Deep convolutional triplet network for phenotypic distance calculation.
result Euclidean phenotypic distances support classical mimicry theory.
Theory explains generalization in deep learning, reducing memorization and improving performance.
problem Understanding and improving generalization in deep learning models.
method Developed a non-asymptotic theory using the empirical neural tangent kernel.
result Generalization is possible even when the kernel evolves significantly, with coherent signal accumulation and noise suppression.
Many theories of deep learning have shown that a deep network can require dramatically fewer resources to represent a given function compared to a shallow network. But a question remains: can these efficient representations be learned using current deep learning techniques? In this work, we test whether standard deep l…
Algorithm constructs confidence sets for deep neural networks with PAC guarantees.
problem Ensuring reliable predictions for deep neural networks with high confidence.
method Combines calibrated prediction and learning theory bounds.
result Constructs PAC confidence sets for various deep models.
Abstract: Surveying connections between ML and Control Theory.
problem Addressing the intersection of Machine Learning and Control Theory.
method Develops connections through reinforcement learning, supervised learning, deep learning, and stochastic gradient descent.
result Machine Learning and Control Theory are interconnected, with ML solving large control problems and Control Theory providing tools for ML.
This paper extends RMT for deep learning models beyond eigenvalues.
problem Challenges in high-dimensional, overparameterized ML models.
method Introduces High-dimensional Equivalent to analyze nonlinear models.
result Unified understanding of training and generalization in deep learning.
Study of infinitely deep but narrow neural networks using NTK theory.
problem Analyzing the role of depth in deep learning with overparameterized networks.
method Infinite-depth limit analysis of MLP and CNN using Neural Tangent Kernel (NTK) theory.
result Established trainability guarantee for infinitely deep but narrow neural networks.
In this paper we develop a statistical theory and an implementation of deep learning models. We show that an elegant variable splitting scheme for the alternating direction method of multipliers optimises a deep learning objective. We allow for non-smooth non-convex regularisation penalties to induce sparsity in parame…
A framework for analyzing regularizers to ensure trustworthy theory-driven model estimation.
problem Uncertain choice of regularizers can compromise the interpretability of deep grey-box models.
method Adapting neural net architecture and training objective to analyze regularizer behavior empirically.
result Empirical analysis of regularizers helps in making a justified choice for trustworthy theory-driven model estimation.
New insights into continual learning for deep models, showing convergence issues but local linear solutions.
problem Challenges in continual learning for homogeneous deep models.
method Sequential projections onto task margin sets, leveraging nonconvex projection theory.
result Local linear convergence under certain conditions for homogeneous deep networks.
The paper provides statistical guarantees for sparse deep learning.
problem Understanding the potential and limitations of sparse deep learning.
method Develops statistical guarantees for different types of sparsity in sparse deep learning.
result Statistical guarantees for sparse deep learning with mild dependence on network widths and depths.
New framework explains deep neural networks using variational spline theory.
problem Understanding functions learned by deep neural networks.
method Developed a variational framework and function space.
result Deep ReLU networks are solutions to regularized data fitting problems over the proposed function space.
Deep learning enhances Hamiltonian Monte Carlo for sampling gauge field configurations.
problem Sampling from complex gauge field topologies efficiently.
method Stacked neural networks to generalize Hamiltonian Monte Carlo.
result Significantly reduces computational cost for generating gauge field configurations.