Stochastic approach improves neural network training for kinetic simulations.
problem Training neural networks under physical constraints in kinetic fusion simulations.
method Stochastic augmented Lagrangian approach using pyTorch.
result Higher model prediction accuracy achieved compared to fixed penalty method.
Stochastic gradient methods converge for training wide PINNs.
problem Convergence of stochastic gradient descent in training over-parameterized PINNs.
method Established linear convergence of stochastic gradient descent/flow in training over-parameterized two-layer PINNs.
result Linear convergence with high probability for general activation functions.
Langevin algorithms enhance training of deep neural networks for stochastic control problems.
problem Training acceleration for deep neural networks in stochastic control problems.
method Application of Langevin algorithms to minimize the loss of deep neural networks in stochastic control problems.
result Langevin algorithms improve training on various stochastic control problems.
Stochastic binary hidden units in a multi-layer perceptron (MLP) network give at least three potential benefits when compared to deterministic MLP networks. (1) They allow to learn one-to-many type of mappings. (2) They can be used in structured prediction problems, where modeling the internal structure of the output i…
Dropout and similar stochastic neural network regularization methods are often interpreted as implicitly averaging over a large ensemble of models. We propose STE (stochastically trained ensemble) layers, which enhance the averaging properties of such methods by training an ensemble of weight matrices with stochastic r…
Wide stochastic networks show Gaussian behavior and improve training with PAC-Bayesian methods.
problem Analyzing and training over-parameterised neural networks with large width.
method Establishing Gaussian behavior for a stochastic architecture, applying PAC-Bayesian training.
result PAC-Bayesian training on large but finite-width networks outperforms standard methods.
Stochastic LWTA networks resist adversarial attacks while maintaining accuracy.
problem Adversarial robustness of neural networks.
method Replaced ReLU with stochastic LWTA activations, trained with Variational Bayesian and PGD.
result Stochastic LWTA networks achieve state-of-the-art robustness against adversarial attacks.
During the last few years, significant attention has been paid to the stochastic training of artificial neural networks, which is known as an effective regularization approach that helps improve the generalization capability of trained models. In this work, the method of modified equations is applied to show that the r…
Scalable Gaussian process models trained with unbiased stochastic ELBO.
problem Training large capacity Gaussian process models on huge datasets.
method Unbiased stochastic variational inference for scalable GPs.
result Accurate inference on large datasets with up to 10 million basis functions.
Regularized SB process speeds up generative modeling.
problem Slow sampling and training times in SB-based models.
method Regularization terms to reduce timesteps and training time.
result Faster sampling speed for generative modeling.
Stochastic trace estimation with tensor train random vectors
problem Stochastic trace estimation for large-scale matrices
method Gaussian random tensor train vectors
result Median-of-means variant achieves dimension-independent guarantees
SMGD trains low-bit neural networks with memory constraints.
problem Training large neural networks with limited memory.
method Stochastic Markov Gradient Descent (SMGD).
result Encouraging numerical results and theoretical guarantees.
What enables Stochastic Gradient Descent (SGD) to achieve better generalization than Gradient Descent (GD) in Neural Network training? This question has attracted much attention. In this paper, we study the distribution of the Stochastic Gradient Noise (SGN) vectors during the training. We observe that for batch sizes …
Surrogate models speed up RL training in dynamic systems.
problem High computational cost of high-fidelity simulations.
method Developed and tested surrogate models for RL training.
result Surrogate models can significantly accelerate RL training.
SPEQ improves quantized neural networks by stochastic precision sharing and cosine similarity loss.
problem Improving quantized deep neural networks for edge devices.
method SPEQ combines stochastic precision sharing and cosine similarity loss for knowledge distillation.
result SPEQ outperforms existing methods in various tasks.
We apply stochastic average gradient (SAG) algorithms for training conditional random fields (CRFs). We describe a practical implementation that uses structure in the CRF gradient to reduce the memory requirement of this linearly-convergent stochastic gradient method, propose a non-uniform sampling scheme that substant…
New algorithm optimizes PAC-Bayes bound without surrogate loss.
problem Mismatch between optimisation objective and generalisation bound in stochastic neural networks.
method Proposes a novel training algorithm that optimizes the PAC-Bayesian bound directly.
result Empirical results show improved performance over existing PAC-Bayesian training methods.
SGD fails to converge for deep ReLU networks with limited random initializations.
problem SGD convergence in deep neural networks with limited random initializations.
method Analysis of four discretization parameters: network architecture, training data, gradient steps, and random initializations.
result SGD fails to converge for ReLU networks with depth much larger than width.
Adaptive methods optimize machine learning training faster.
problem Non-adaptive stochastic optimization requires tuning for each application.
method Develop adaptive stochastic optimization methods.
result Adaptive methods offer computational savings for large-scale systems.
We study the problem of training deep neural networks with Rectified Linear Unit (ReLU) activation function using gradient descent and stochastic gradient descent. In particular, we study the binary classification problem and show that for a broad family of loss functions, with proper random weight initialization, both…
New analysis reveals batch size effects on stochastic conditional gradient methods.
problem Understanding the role of batch size in stochastic conditional gradient methods.
method Deriving a new analysis focusing on momentum-based stochastic conditional gradient algorithms (e.g., Scion).
result Increasing batch size initially improves optimization accuracy but can degrade performance beyond a critical threshold.
Stochastic normalizing flows use SDEs for efficient training and sampling.
problem Efficient maximum likelihood estimation and variational inference.
method Continuous normalizing flows extended with stochastic differential equations (SDEs) and rough path theory.
result Stochastic normalizing flows enable efficient training and sampling from complex distributions.
Deep learning solves complex stochastic control with jumps.
problem Solving high-dimensional stochastic control tasks with jumps.
method Model-based approach using two neural networks, iteratively trained with objectives derived from the Hamilton-Jacobi-Bellman equation.
result Demonstrates effectiveness in solving complex high-dimensional stochastic control tasks.
A new training method uses multilevel minimization for machine learning.
problem Training machine learning models with high variance and low efficiency.
method Constructs a multilevel hierarchy by reducing sample size and internally trains surrogate models with fewer samples.
result The multilevel method enhances model training efficiency compared to subsampled Newton's and variance reduction methods.
Paper examines the structure of stochastic gradients in deep learning.
problem Exploring the structure and heavy tails of stochastic gradients in deep learning.
method Conducted formal statistical tests on stochastic gradients and gradient noise.
result Stochastic gradients and gradient noise do not exhibit power-law heavy tails, but their covariance spectra do.
This paper uses SDEs to analyze GANs training and long-run behavior.
problem Understanding the training process and long-run behavior of GANs.
method Established SDE approximations for GANs training and analyzed long-run behavior via invariant measures.
result The long-run behavior of GANs training can be studied via the invariant measures of its SDE approximations.
Stochastic neural networks with infinite width become deterministic, reducing training variance.
problem Understanding how stochasticity in neural networks affects learning and regularization.
method Theoretical analysis of stochastic neural networks with infinite width.
result As the width of an optimized stochastic neural network increases, its predictive variance on the training set decreases to zero.
Deep ensembles effectively capture epistemic uncertainty through training stochasticity, providing a frequentist perspective.
problem Understanding and quantifying epistemic uncertainty in machine learning models.
method Bootstrap-based estimator and decomposition of deep ensembles into data variability and training stochasticity.
result Deep ensembles primarily capture training stochasticity, explaining their effectiveness in quantifying epistemic uncertainty.
Empirical study shows GANs overfit and drop modes when training is deterministic.
problem Understanding overfitting and mode drop in GAN training.
method Empirical analysis of GAN training with and without stochasticity.
result GANs overfit and drop modes when training is deterministic.
Stochastic momentum methods have been widely adopted in training deep neural networks. However, their theoretical analysis of convergence of the training objective and the generalization error for prediction is still under-explored. This paper aims to bridge the gap between practice and theory by analyzing the stochast…
Training-free model learns SDE dynamics without training, accelerating parameter studies.
problem High computational cost of simulating parameter-dependent SDEs.
method Training-free conditional diffusion model with joint kernel-weighted Monte Carlo estimator.
result Accurate approximation of conditional distributions across varying parameter values.
Stochastic Gradient Descent introduces noise in training, affecting model decision boundaries.
problem Understanding the impact of noise in SGD on model decision boundaries.
method Characterized SGD and persistent SGD dynamics in a neural network model, measuring noise magnitude in both under- and over-parametrized regimes.
result Noisier algorithms lead to wider decision boundaries in constraint satisfaction problems.
Generative models for complex stochastic dynamics using adversarial learning.
problem Data-driven modeling of multistep stochastic dynamics.
method Adversarial learning with GANs and MMD for stable model classes.
result Stable generative models for long-time prediction and stochastic systems.
This paper studies the performance of a recently proposed preconditioned stochastic gradient descent (PSGD) algorithm on recurrent neural network (RNN) training. PSGD adaptively estimates a preconditioner to accelerate gradient descent, and is designed to be simple, general and easy to use, as stochastic gradient desce…
Normalization layers improve the accuracy of Differentially Private training of deep neural networks.
problem Reduced accuracy in deep neural networks with Differentially Private training.
method Proposed a novel method for integrating batch normalization with Differentially Private Stochastic Gradient Descent (DPSGD) without additional privacy loss.
result Training deeper networks with better utility-privacy trade-off is possible.
Paper introduces FDM for efficient training of Neural SDEs.
problem Training Neural SDEs using existing methods is computationally expensive and unstable.
method Developed a novel scoring rule called Finite Dimensional Matching (FDM) to bypass signature kernels and reduce training complexity.
result FDM achieves superior performance in terms of computational efficiency and generative quality.
New method improves training stochastic neural networks with tighter guarantees.
problem Training stochastic neural networks with provable guarantees.
method Developed partially-aggregated estimators and reformulated PAC-Bayesian bounds.
result Derives a differentiable objective leading to tighter generalisation guarantees.
Study models deep learning training dynamics using locally elastic SDEs to reveal feature separability.
problem Understanding how deep learning models separate features from different classes during training.
method Modeling deep learning training using locally elastic SDEs with a drift term reflecting backpropagation impact.
result Local elasticity in SDEs leads to linear separability of features, resulting in vanishing training loss.
Framework for training stochastic spiking neural networks with rough signals.
problem Training stochastic spiking neural networks with noisy spike timing and dynamics.
method Rough path theory and signature kernels for gradient computation.
result Pathwise gradients of SSNNs' trajectories and event times exist and satisfy a recursive relation.
We show that parametric models trained by a stochastic gradient method (SGM) with few iterations have vanishing generalization error. We prove our results by arguing that SGM is algorithmically stable in the sense of Bousquet and Elisseeff. Our analysis only employs elementary tools from convex and continuous optimizat…
Proposes SWA for adversarial training to improve model robustness.
problem Overfitting in adversarial training due to insufficient sample complexity.
method Adversarial training with Stochastic Weight Average (SWA).
result Improves model robustness on CIFAR-10, CIFAR-100, and SVHN datasets.
Predict training time without training using a linearized model.
problem Predicting the training time of a deep network without actual training.
method Approximate training dynamics with a linearized model and solve a low-dimensional SDE in function space.
result Predict training time within 20% error margin and 30-45x reduction in cost.
Develops Bayesian approach for end-to-end learning in stochastic optimization.
problem Stochastic optimization problems under uncertainty.
method Bayesian interpretation and new end-to-end learning algorithms.
result Improved decision maps for empirical risk minimization and distributionally robust optimization.
SDE Matching eliminates simulation for training Latent SDEs, achieving similar performance.
problem Training Latent SDEs with adjoint sensitivity methods is computationally expensive and limited.
method SDE Matching, inspired by Score- and Flow Matching, eliminates simulation for training Latent SDEs.
result SDE Matching achieves performance comparable to adjoint sensitivity methods while reducing computational complexity.
This work improves neural likelihood surrogates for stochastic models with a score-augmented loss.
problem Efficient parameter inference for stochastic models with computationally expensive likelihood functions.
method Score-augmented loss function for neural network likelihood surrogates.
result Improves surrogate quality at a lower computational cost compared to generating more data.
SNGM improves large-batch training accuracy.
problem Improving generalization in large-batch training.
method Stochastic Normalized Gradient Descent with Momentum.
result SNGM achieves better test accuracy than MSGD and other large-batch methods.
We propose Stochastic Weight Averaging in Parallel (SWAP), an algorithm to accelerate DNN training. Our algorithm uses large mini-batches to compute an approximate solution quickly and then refines it by averaging the weights of multiple models computed independently and in parallel. The resulting models generalize equ…
A machine learning framework predicts self-induced stochastic resonance in neurons.
problem Predicting coherent oscillations in slow-fast excitable systems driven by noise.
method Physics-informed machine learning with a Noise-Augmented State Predictor architecture and Kramers' escape theory constraints.
result Trained PINN accurately predicts spike-train coherence on noise intensity, excitability, and timescale separation.