Large-batch stochastic gradient descent (SGD) is widely used for training in distributed deep learning because of its training-time efficiency, however, extremely large-batch SGD leads to poor generalization and easily converges to sharp minima, which prevents naive large-scale data-parallel SGD (DP-SGD) from convergin…
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.
168,742 papers · 148 categories
Trend · papers per month
48 results for “large noise scales”
Proposes a progressive label correction method for feature-dependent label noise.
problem Real-world large-scale datasets often suffer from heterogeneous, feature-dependent label noise.
method A progressive label correction algorithm that iteratively refines the model.
result A classifier trained with this strategy converges to be consistent with the Bayes classifier for various noise patterns.
Collecting large-scale data with clean labels for supervised training of neural networks is practically challenging. Although noisy labels are usually cheap to acquire, existing methods suffer a lot from label noise. This paper targets at the challenge of robust training at high label noise regimes. The key insight to …
Improved image classification accuracy with a probabilistic model of label noise.
problem Noisy labels in large-scale image classification datasets.
method A probabilistic model using a multivariate Normal distribution on the final hidden layer of a neural network, capturing input-dependent label noise.
result Significantly improved accuracy on various datasets compared to standard methods.
The Effect of Network Width on Stochastic Gradient Descent and Generalization: an Empirical Studycs.LG
We investigate how the final parameters found by stochastic gradient descent are influenced by over-parameterization. We generate families of models by increasing the number of channels in a base network, and then perform a large hyper-parameter search to study how the test error depends on learning rate, batch size, a…
Non-Asymptotic Analysis of Tangent Space Perturbationphysics.data-an
Constructing an efficient parameterization of a large, noisy data set of points lying close to a smooth manifold in high dimension remains a fundamental problem. One approach consists in recovering a local parameterization using the local tangent plane. Principal component analysis (PCA) is often the tool of choice, as…
Generative models improve for multiscale scientific data with new noise and interpolation techniques.
problem Numerical challenges in generating high-fidelity samples for multiscale scientific data.
method Design of noise distributions and interpolation schedules in function space to ensure Lipschitz regularity and finite noise roughness.
result Scale-adaptive noise and interpolation schedules improve numerical efficiency and fidelity of generated samples.
New method improves robustness of large models without sacrificing accuracy.
problem Improving robustness of large pre-trained models without accuracy loss.
method Multi-scale diffusion denoised smoothing, selectively applying smoothing at multiple noise scales.
result Strong certified robustness at high noise levels with accuracy close to non-smoothed classifiers.
This paper enhances privacy-preserving randomized power method for large datasets.
problem Privacy issues in applying randomized power method to large datasets containing personal information.
method Proposes enhanced privacy-preserving variants of the randomized power method, including a variant with reduced noise and a decentralized framework.
result Tighter convergence bounds and empirical comparisons with previous work in real recommendation datasets.
New method reduces memorization in diffusion models without sacrificing image quality.
problem Diffusion models often memorize training data, especially with small datasets.
method Train models using noisy data at large noise scales to reduce memorization.
result Significant reduction in memorization without compromising image quality.
Label noise in SGD helps converge to flatter minima.
problem Improving generalization in overparametrized models.
method Analyzes SGD with label noise, showing convergence to regularized minima.
result SGD with label noise converges to flatter minima, improving generalization.
Study on the noise in SGD minibatches near local minima.
problem Understanding the noise in SGD minibatches near local minima.
method Detailed analysis of SGD noise in linear regression and derivation of a general formula for different types of minima.
result Provides insight into the stability of training neural networks and suggests large learning rates can help generalization.
Unified framework for discrete diffusion modeling with flexible noising processes.
problem Efficient modeling of large discrete state spaces with arbitrary corruption dynamics.
method Generalized Discrete Diffusion from Snapshots (GDDS) framework that supports uniformization for fast noising and snapshot-based ELBO for reverse process.
result GDDS outperforms existing discrete diffusion methods in training efficiency and generation quality.
Stochastic Gradient Langevin Dynamics (SGLD) is a sampling scheme for Bayesian modeling adapted to large datasets and models. SGLD relies on the injection of Gaussian Noise at each step of a Stochastic Gradient Descent (SGD) update. In this scheme, every component in the noise vector is independent and has the same sca…
DiSK improves DP optimizers by simplifying Kalman filtering for better performance.
problem Performance drop of DP optimizers in large-scale training due to noise injection.
method DiSK uses Kalman filtering to denoise privatized gradients and refine gradient estimations.
result DiSK achieves significant performance improvements over standard DP optimizers in large-scale training.
One way to avoid overfitting in machine learning is to use model parameters distributed according to a Bayesian posterior given the data, rather than the maximum likelihood estimator. Stochastic gradient Langevin dynamics (SGLD) is one algorithm to approximate such Bayesian posteriors for large models and datasets. SGL…
Paper shows noisy labels can improve PLR variable selection.
problem Variable selection in PLR is challenging due to noisy labels.
method Proposes a novel ADMM-based algorithm to fuse noisy labels.
result Fused noisy labels improve PLR performance in estimation and classification.
SAP: Corrective Machine Unlearning with Scaled Activation Projection for Label Noise Robustnesscs.LG
SAP corrects model for label noise by identifying and removing noisy samples.
problem Label corruption degrades model performance; acquiring perfect labels is costly.
method SAP uses SVD to identify and project model weights onto a clean activation space.
result SAP improves model generalization by up to 6% on CIFAR dataset with 25% synthetic corruption.
Proposes a text perturbation method using a Mahalanobis metric to balance privacy and utility.
problem Low utility of text analysis when using spherical noise for privacy-preserving text embedding.
method Regularized Mahalanobis metric to add elliptical noise, accounting for embedding space density.
result Improves privacy statistics while maintaining utility, outperforming Laplace mechanism.
Classical scaling is shown to be optimal under various noisy conditions.
problem Consistency of classical scaling under general noise conditions.
method Established using finite fourth moments of noise, derived convergence rates, and matching minimax lower bounds.
result Classical scaling achieves minimax optimality in recovering true configuration from noisy dissimilarities.
We study whether a neural network optimizes to the same, linearly connected minimum under different samples of SGD noise (e.g., random data order and augmentation). We find that standard vision models become stable to SGD noise in this way early in training. From then on, the outcome of optimization is determined to a …
Private Topic Modelingstat.ML
We develop a privatised stochastic variational inference method for Latent Dirichlet Allocation (LDA). The iterative nature of stochastic variational inference presents challenges: multiple iterations are required to obtain accurate posterior distributions, yet each iteration increases the amount of noise that must be …
This paper studies large-scale dynamical networks where the current state of the system is a linear transformation of the previous state, contaminated by a multivariate Gaussian noise. Examples include stock markets, human brains and gene regulatory networks. We introduce a transition matrix to describe the evolution, …
Paper tackles label noise in large datasets, purifying noisy data with a nonparametric framework.
problem Label noise in large-scale datasets with coarse labels.
method Develops a model-agnostic nonparametric framework for classification.
result Framework purifies noisy data using a small clean dataset and manages ambiguous samples.
NCVis speeds up data visualization for large datasets.
problem Performance issues in t-SNE for large datasets.
method Noise contrastive estimation for scalable visualization.
result NCVis outperforms state-of-the-art techniques in speed and quality.
Sparse Convex Biclusteringstat.ML
Sparse Convex Biclustering improves accuracy and robustness in high-dimensional datasets.
problem Challenges in clustering rows and columns of large-scale datasets due to noise and computational complexity.
method Sparse Convex Biclustering (SpaCoBi) using convex optimization and stability-based tuning.
result Significantly outperforms state-of-the-art methods in accuracy for high-dimensional datasets.
Stochastic gradient descent with noise of machine learning type. Part II: Continuous time analysiscs.LG
Continuous-time analysis shows SGD with noise prefers flat minima.
problem Optimizing neural networks using SGD with noise.
method Continuous-time model for SGD with noise analysis.
result Optimization prefers flat minima in certain noise regimes.
In this paper, we present a new practical method for Bayesian learning that can rapidly draw representative samples from complex posterior distributions with multiple isolated modes in the presence of mini-batch noise. This is achieved by simulating a collection of replicas in parallel with different temperatures and p…
Applying Guidance in a Limited Interval Improves Sample and Distribution Quality in Diffusion Modelscs.CV
Improved image quality in diffusion models by limiting guidance to a specific noise level range.
problem Improving image quality in diffusion models with guidance.
method Restricting guidance to a specific noise level range.
result Significantly improved FID score from 1.81 to 1.40 in ImageNet-512.
KFT improves tensor forecasting by incorporating side information.
problem Tensor factorization weaknesses in latent factors.
method Kernel Fried Tensor (KFT) with variational inference.
result Superior performance over LightGBM and FFM.
TCR improves DNN robustness to noisy labels with minimal overhead.
problem Training on noisy labeled datasets degrades DNN generalization.
method TCR combines original labels and previous epoch predictions for regularization.
result TCR consistently enhances DNN robustness to label noise.
Gradient descent with noise converges to a unique optimum in nonconvex matrix factorization.
problem Gradient descent with noise converges to a unique optimum in nonconvex matrix factorization.
method A perturbed form of gradient descent with arbitrary initialization.
result Gradient descent with noise converges to a unique optimum.
AlignFlow improves FGMs by optimizing noise and data alignment.
problem Optimal Transport methods for FGMs are limited by scalability issues.
method Introduces Semi-Discrete Optimal Transport (SDOT) to enhance FGM training.
result AlignFlow scales well to large datasets and model architectures.
Deep neural networks have established as a powerful tool for large scale supervised classification tasks. The state-of-the-art performances of deep neural networks are conditioned to the availability of large number of accurately labeled samples. In practice, collecting large scale accurately labeled datasets is a chal…
The performance of the Lasso is well understood under the assumptions of the standard linear model with homoscedastic noise. However, in several applications, the standard model does not describe the important features of the data. This paper examines how the Lasso performs on a non-standard model that is motivated by …
Differentially private log-location-scale regression models improve privacy in statistical analysis.
problem Ensuring privacy in statistical regression models while maintaining accuracy.
method Integrates differential privacy into LLS regression using the functional mechanism.
result Proposed DP-LLS models satisfy ε-differential privacy and perform well under various conditions.
The paper studies how noise synchronizes tokens in deep transformer models.
problem Understanding synchronization in deep learning models with noise.
method Proves convergence to a stochastic particle system and identifies the limiting SDE.
result The limiting model displays synchronization by noise and exponential dissipation of interaction energy.
Paper introduces robust kernel ridge regression using Cauchy loss for handling various noise types.
problem Developing robust regression methods for noisy data.
method Introduces kernel Cauchy ridge regressor (KCRR) using Cauchy loss function.
result Establishes almost minimax-optimal convergence rate for KCRR in terms of -risk.
This work reveals symmetries in quantum circuits and develops a noise-aware optimization method.
problem Understanding and optimizing the cost landscape of parametrized quantum circuits.
method Analytical proof of symmetries and their resilience to noise, followed by the development of SYMH optimization method.
result Symmetries in PQCs lead to degeneracy in the cost landscape and can be exploited to improve optimization under noise.
SFM resolves small-scale physics challenges in weather data.
problem Challenges in super-resolving small-scale details in physical sciences like weather.
method Encoding inputs to a latent base distribution, flow matching for stochastic details, adaptive noise scaling.
result SFM framework significantly outperforms existing methods.
The gradient noise of SGD is considered to play a central role in the observed strong generalization abilities of deep learning. While past studies confirm that the magnitude and the covariance structure of gradient noise are critical for regularization, it remains unclear whether or not the class of noise distribution…
Bayesian inference promises to ground and improve the performance of deep neural networks. It promises to be robust to overfitting, to simplify the training procedure and the space of hyperparameters, and to provide a calibrated measure of uncertainty that can enhance decision making, agent exploration and prediction f…
Scale-adaptive and robust intrinsic dimension estimation via optimal neighbourhood identificationstat.ML
This work introduces a protocol to automatically select the correct range of scales for meaningful Intrinsic Dimension estimation.
problem The Intrinsic Dimension (ID) varies with scale in real-world datasets, leading to erroneous results.
method The protocol selects the correct range of scales by ensuring constant density of data points.
result The method provides a robust and scale-adaptive approach to estimating meaningful Intrinsic Dimension.
Scaled sparse linear regression jointly estimates the regression coefficients and noise level in a linear model. It chooses an equilibrium with a sparse regression method by iteratively estimating the noise level via the mean residual square and scaling the penalty in proportion to the estimated noise level. The iterat…
Initialization Matters: On the Benign Overfitting of Two-Layer ReLU CNN with Fully Trainable Layerscs.LG
Two-layer CNNs can overfit well if initialized correctly.
problem Understanding the conditions for benign overfitting in over-parameterized CNNs.
method Extending analysis to fully trainable two-layer CNNs, examining initialization scaling effects.
result Initialization scaling of the output layer is crucial; large scales lead to fixed output behavior, small scales to complex interactions.
Corrupting the input and hidden layers of deep neural networks (DNNs) with multiplicative noise, often drawn from the Bernoulli distribution (or 'dropout'), provides regularization that has significantly contributed to deep learning's success. However, understanding how multiplicative corruptions prevent overfitting ha…
Stochastic Approximation with Markov Noise: Analysis and applications in reinforcement learningcs.LG
We analyze stochastic approximation with Markov noise for reinforcement learning.
problem Asymptotic convergence analysis of stochastic approximation with controlled Markov noise.
method Relating to limiting differential inclusions and ergodic occupation measures.
result First informative error bounds on function approximation for policy evaluation.
eDCF estimates intrinsic dimension using local connectivity.
problem Challenges in estimating intrinsic dimension due to scale dependence.
method eDCF: a novel, scalable, and parallelizable method based on Connectivity Factor (CF).
result eDCF consistently matches leading estimators with comparable MAE and higher exact intrinsic dimension match rates.