Scaling laws in linear regression explain model performance improvements with size and data.
problem Disagreement between empirical neural scaling laws and conventional wisdom on variance error.
method Infinite dimensional linear regression setup, one-pass SGD, Gaussian prior, power-law spectrum.
result Variance error is dominated by other errors, disappearing from the bound due to SGD's implicit regularization.
Study on learning properties of scale-dependent kernels controlling stability and error.
problem Understanding the learning properties of scale-dependent kernels in nonparametric ridge-less least squares.
method Combines probabilistic results with interpolation theory to analyze stability and error.
result Different regimes of learning error depending on sample size and data dimension.
A new method for reinforcement learning scales errors without tuning.
problem Error scaling varies across reinforcement learning tasks and stages.
method A simple scaling mechanism for temporal-difference learning.
result The method effectively mitigates interference between learning tasks.
The paper explores neural scaling laws for deep operator networks, offering a theoretical foundation.
problem Understanding neural scaling laws in deep operator networks.
method Theoretical analysis of approximation and generalization errors.
result Established a theoretical framework to quantify neural scaling laws for deep operator networks.
Study rates of convergence for approximate solutions to linear ill-posed problems in Hilbert scales.
problem Linear ill-posed inverse problems with noisy data.
method Approximate reconstructions from random noisy data using regularization schemes in Hilbert scale.
result Explicitly established error bounds for smooth regression functions.
The dependency of the generalization error of neural networks on model and dataset size is of critical importance both in practice and for understanding the theory of neural networks. Nevertheless, the functional form of this dependency remains elusive. In this work, we present a functional form which approximates well…
Recent works have derived non-asymptotic upper bounds for convergence of underdamped Langevin MCMC. We revisit these bound and consider introducing scaling terms in the underlying underdamped Langevin equation. In particular, we provide conditions under which an appropriate scaling allows to improve the error bounds in…
This research examines how the error rate of nearest neighbor classifiers varies with dataset size.
problem The scaling of classification error rates with dataset size is not uniform.
method Theoretical analysis of nearest neighbor classifiers, focusing on early and late phases of dataset size.
result The error rate of nearest neighbor classifiers can have fine-grained rates depending on the dataset size and data distribution.
In theoretical analysis of deep learning, discovering which features of deep learning lead to good performance is an important task. In this paper, using the framework for analyzing the generalization error developed in Suzuki (2018), we derive a fast learning rate for deep neural networks with more general activation …
Pruned neural networks' error scales predictably with architecture and task.
problem Understanding the predictability of pruning across different scales and architectures.
method Functionally approximated the error of pruned networks, showing it is predictable in terms of invariant tying width, depth, and pruning level.
result The error of pruned networks follows a scaling law with interpretable coefficients that depend on architecture and task.
Improved scaling laws in linear regression using data reuse.
problem Sustainability of neural scaling laws when running out of new data.
method Data reuse in multi-pass stochastic gradient descent (multi-pass SGD) for M-dimensional linear models trained on N data with sketched features. result Multi-pass SGD achieves a test error of Θ(M1−b+L(1−b)/a) with L>N, improving scaling laws in data-constrained regimes. We present a new method for high-dimensional linear regression when a scale parameter of the additive errors is unknown. The proposed estimator is based on a penalized Huber M-estimator, for which theoretical results on estimation error have recently been proposed in high-dimensional statistics literature. However, t…
New pruning method breaks power law scaling, potentially reducing error to exponential.
problem Improving neural network performance through scaling alone is costly.
method Developed a new data pruning metric to break power law scaling.
result Pruned datasets show better than power law scaling on various image datasets.
Study how generalization scales with model size and data in quadratic neural networks.
problem Understanding how generalization scales with model size and data in quadratic neural networks.
method Analyzed ℓ2-regularized empirical test error minimization in a quadratic two-layer network with finite-sample setting and structured data. result Revealed a phase diagram with distinct scaling regimes as the number of parameters varies, showing data-dependent power laws controlled by spectral structure of the target.
Deep learning (DL) creates impactful advances following a virtuous recipe: model architecture search, creating large training data sets, and scaling computation. It is widely believed that growing training sets and models should improve accuracy and result in better products. As DL application domains grow, we would li…
This work studies scaling laws for low-precision training in high-dimensional linear regression.
problem Optimizing trade-off between model quality and training costs in high-dimensional linear regression.
method Theoretical study of scaling laws for low-precision training within a high-dimensional sketched linear regression framework, analyzing multiplicative and additive quantization.
result Multiplicative quantization maintains full-precision model size, while additive quantization reduces effective model size.
Derives error bounds for stochastic iterative algorithms using Stein's method.
problem Bounding errors in stochastic iterative algorithms like SGD and SGLD.
method Uses infinite-dimensional Stein's method of exchangeable pairs to derive functional approximation error bounds.
result Establishes non-asymptotic error bounds for algorithm sample paths and variance of iterate averages.
Applications such as weather forecasting and personalized medicine demand models that output calibrated probability estimates---those representative of the true likelihood of a prediction. Most models are not calibrated out of the box but are recalibrated by post-processing model outputs. We find in this work that popu…
This work analyzes neural scaling laws using power-law data spectra and derives analytical expressions for generalization error.
problem Understanding how neural network performance scales with key factors like data size and model complexity.
method Statistical mechanics techniques applied to one-pass stochastic gradient descent in a student-teacher framework.
result Derivation of analytical expressions for generalization error under power-law data spectra and identification of conditions for power-law scaling.
Inverse depth scaling found in LLMs due to similar layers averaging error.
problem Understanding how depth affects loss in large language models.
method Analysis of LLMs and toy residual networks.
result Loss scales inversely proportional to depth in LLMs.
The theory of quantum computation can be constructed from the abstract study of anyonic systems. In mathematical terms, these are unitary topological modular functors. They underlie the Jones polynomial and arise in Witten-Chern-Simons theory. The braiding and fusion of anyonic excitations in quantum Hall electron liqu…
Model uncertainty obtained by variational Bayesian inference with Monte Carlo dropout is prone to miscalibration. The uncertainty does not represent the model error well. In this paper, temperature scaling is extended to dropout variational inference to calibrate model uncertainty. Expected uncertainty calibration erro…
The paper analyzes learning curves for kernel ridge regression with dot-product kernels.
problem Understanding the learning curves for different scaling regimes of data and model.
method Precise formulas for mean test error, bias, and variance in the mo∞ with m/dr constant regime. result A peak in the learning curve at m≈dr/r! for any integer r. New principles needed for scaling large language models, challenging traditional regularization methods.
problem The shift from generalization to scaling in machine learning requires new guiding principles.
method Examining the effectiveness of traditional regularization methods in the scaling-centric era.
result Traditional principles of regularization may not generalize to larger scales, highlighting new phenomena like scaling law crossover.
Improved 2-bit covariance estimator with reduced operator norm error and no tuning needed.
problem Improving 2-bit covariance estimation with reduced operator norm error and no tuning needed.
method Proposed a new 2-bit covariance matrix estimator using triangular dithering scales.
result Improved operator norm error rate that depends on effective rank of covariance matrix, closing theoretical gap.
New method for scalable inference in large-scale regression models with complex error structures.
problem Challenges in statistical inference for large-scale regression models with dependent errors.
method Generalized Method of Wavelet Moments with Exogenous variables (GMWMX).
result Statistical validity and scalability of GMWMX for linear models with complex error structures.
The study explains transformer scaling laws using statistical and approximation theories.
problem Understanding why transformer scaling laws exist for large models trained on low-dimensional data.
method Established statistical estimation and mathematical approximation theories for transformers on low-dimensional manifolds.
result Predicted a power law between generalization error and model and data sizes, with power depending on intrinsic data dimension.
This work shows how approximate reward models can significantly improve inference-time scaling.
problem Improving the efficiency of inference for large language models.
method Identifying the Bellman error of approximate reward models and using Sequential Monte Carlo (SMC) for inference.
result Approximate reward models can reduce computational complexity from exponential to polynomial in T. Sharp theory of neural network scaling laws for hierarchical targets.
problem Learning hierarchical multi-index models in neural networks.
method Sharp information-theoretic scaling laws derived for two-layer neural networks.
result Optimal rates achieved by a simple spectral estimator.
New analysis shows Local SGD can achieve error scaling with only fixed number of communications.
problem Speeding up SGD by parallelizing across multiple workers with reduced communication overhead.
method Proposed and analyzed Local SGD method with a fixed number of communications independent of the number of steps.
result Achieves an error scaling as 1/(nT) with only a fixed number of communications (Ω(n)).
Improved machine learning models outperform their simpler counterparts by using imperfect labels.
problem Improving model performance using imperfect labels.
method Random feature ridge regression (RFRR) with a deterministic equivalent for excess test error.
result The student model can outperform the teacher model regardless of the teacher's scaling law, achieving the minimax optimal rate.
Proposes a method to predict time series data using scale information.
problem Sequential data exhibits different behavior at different scales.
method Scale-based inference using k-nearest neighbors and sequential prediction. result Asymptotically, the algorithm produces the best prediction error.
Develops a simple model to understand learning curves for arbitrary power laws.
problem Lack of theoretical understanding of scaling laws in machine learning.
method Analyzes a toy model to determine if learning curves are universal or depend on data distribution.
result Determines that learning curves can exhibit n−β for arbitrary power β>0. Paper analyzes convergence of two time-scale stochastic approximation using martingale approach.
problem Analyzing convergence of two time-scale stochastic approximation algorithms.
method Uses martingale approach to establish convergence conditions and rates.
result Establishes different rates of convergence for fast and slow subsystems.
We investigate finite-time decoupled convergence in nonlinear two-time-scale stochastic approximation.
problem Achieving decoupled convergence in nonlinear two-time-scale stochastic approximation.
method Nested local linearity assumption, suitable step size selection, convergence analysis of matrix cross term, fourth-order moment convergence rates.
result Finite-time decoupled convergence rates can be achieved in nonlinear two-time-scale stochastic approximation with proper step size selection.
Study on distributed linear regression performance, focusing on generalization error.
problem Performance of distributed learning in large-scale linear regression.
method Statistical learning approach, focusing on generalization error.
result Generalization error of distributed solution can be higher than centralized solution.
Study precise sample covariance error for Gaussian centered data.
problem Precise characterization of sample covariance error for Gaussian data.
method Developed a Random Duality Theory (RDT) framework to determine upper and lower bounds.
result Upper and lower bounds match in large-dimensional contexts, matching the spectral norm's limiting value.
State space models (SSMs) provide a flexible framework for modeling complex time series via a latent stochastic process. Inference for nonlinear, non-Gaussian SSMs is often tackled with particle methods that do not scale well to long time series. The challenge is two-fold: not only do computations scale linearly with t…
Gradient descent performs well on weakly convex losses, offering generalization guarantees.
problem Learning with weakly convex losses using gradient descent.
method Analyzing the stability of gradient descent through the smallest eigenvalue of the Hessian.
result Generalization error bounds hold under a wider range of step sizes.
Boosting theory explains why multi-scale GNNs work.
problem Over-smoothing in graph neural networks.
method Gradient boosting and transductive learning analysis.
result Test error bound decreases with more node aggregations.
SGLDiff approximates Bayesian posterior distributions with subsampling error.
problem Approximating Bayesian posterior distributions in large-scale data settings.
method Stochastic Gradient Langevin Diffusion (SGLDiff) with subsampling.
result The Wasserstein distance between the posterior and SGLDiff's limiting distribution is bounded by a fractional power of the mean waiting time.
Neural networks, a central tool in machine learning, have demonstrated remarkable, high fidelity performance on image recognition and classification tasks. These successes evince an ability to accurately represent high dimensional functions, but rigorous results about the approximation error of neural networks after tr…
Improved bounds for non-linear SA with fast convergence.
problem Stochastic approximation with non-linear mappings and multiple time scales.
method Mean squared error bounds with O(1/k) rate for contractive mappings. result First O(1/k) rate for non-linear two-time-scale SA without additional smoothness assumptions. Variant of mSSA improves time series prediction error.
problem Improve prediction error in multivariate time series.
method Introduce spatio-temporal factor model, establish prediction error scaling.
result Prediction error scales as 1 / √(min(N, T)T).
New estimators for intrinsic dimension and Wasserstein distance improve OT accuracy.
problem Intrinsic dimension estimation and Wasserstein distance estimation in large-scale OT.
method Introduces novel estimators for intrinsic dimension and Wasserstein distance.
result Simple, tuning-free estimator of OT and fast intrinsic dimension estimator.
Paper studies offline RL with linear approx, focusing on inherent Bellman error.
problem Offline RL with linear approx, focusing on inherent Bellman error.
method Algorithm that succeeds under single-policy coverage condition, leveraging inherent Bellman error.
result Algorithm yields first known guarantee under single-policy coverage, even for linear Bellman completeness.
Deep learning improves image reconstruction, but scaling up training sets doesn't significantly boost performance.
problem Understanding the impact of training set size on deep learning image reconstruction.
method Empirical study and analytical characterization of performance scaling laws.
result Scaling up training set size does not significantly improve reconstruction quality for deep learning.
Study on infinitely-wide CNNs and their adaptability to function spatial scales.
problem Understanding how CNNs efficiently learn high-dimensional functions and their adaptability to function spatial scales.
method Study infinitely-wide deep CNNs in the kernel regime, characterizing their spectrum and using generalisation bounds to prove adaptability.
result Deep CNNs adapt to the spatial scale of the target function, with error decay controlled by the effective dimensionality of function subsets.