Paper proposes a pre-conditioning technique to speed up gradient-descent convergence in distributed linear least-squares problems.
problem Expediting convergence of gradient-descent method for ill-conditioned distributed linear least-squares problems.
method Iterative pre-conditioning technique to improve convergence rate of gradient-descent method.
result Pre-conditioned gradient-descent achieves superlinear convergence for unique solutions and improved linear convergence otherwise.
Paper proposes a pre-conditioning method to speed up gradient descent in multi-agent optimization.
problem Speed up convergence of gradient descent in multi-agent optimization problems.
method Iterative pre-conditioning approach to mitigate the effect of problem conditioning.
result Significant improvement in convergence speed of gradient descent method.
New method speeds up distributed linear regression.
problem Efficiently solve distributed linear regression problems.
method Iteratively Pre-conditioned Stochastic Gradient Descent (IPSG)
result Converges linearly in expectation to the solution.
Modern machine learning-based recognition approaches require large-scale datasets with large number of labelled training images. However, such datasets are inherently difficult and costly to collect and annotate. Hence there is a great and growing interest in automatic dataset collection methods that can leverage the w…
Paper proves PI consensus algorithm converges exponentially under restricted secant inequality.
problem Proving convergence of PI consensus algorithm without convexity.
method Lyapunov theory, restricted secant inequality, rate-matching discretization, local pre-conditioning.
result Exponential convergence of PI consensus algorithm for non-convex functions.
Deeper neural networks learn lower frequency functions faster, according to a new principle.
problem Understanding why deeper learning is faster.
method Fourier analysis and filtering method to separate and analyze the frequency distribution of neural network outputs.
result Deeper hidden layers in neural networks bias towards lower frequency functions during training.
Pre-conditioning is a well-known concept that can significantly improve the convergence of optimization algorithms. For noise-free problems, where good pre-conditioners are not known a priori, iterative linear algebra methods offer one way to efficiently construct them. For the stochastic optimization problems that dom…
This paper accelerates distributed convex optimization by mitigating ill-conditioning issues.
problem Distributed convex optimization with ill-conditioned aggregate cost functions.
method Iterative pre-conditioning technique to improve convergence rate and stability.
result The proposed algorithm converges linearly with improved convergence rate and superlinearly under certain conditions.
Proposes efficient sampling methods for solving linear inverse problems.
problem Solving linear inverse problems with computational efficiency and accuracy.
method Higher-order Langevin diffusion with pre-conditioning and annealing.
result Provable sampling from posterior distributions with accelerated convergence.
Paper proposes a hierarchical approach for early anomaly detection in time series data for critical health events.
problem Early detection of critical health events in intensive care units.
method A layered learning architecture that breaks the problem into pre-conditional and event layers.
result The proposed method outperforms state-of-the-art approaches for critical health episode prediction.
A planning approach learns skills from interactions, balancing exploration and exploitation.
problem Learning robust high-level skills in noisy environments with unknown pre-conditions.
method Formulates skills as high-level policies, learns plans via bandit problems, balances exploration and exploitation.
result A planner capable of learning robust high-level skills in high-dimensional state spaces.
Random Fourier features improve tabular deep learning convergence.
problem Tabular deep learning convergence issues.
method Random Fourier projections as a pre-processing step, projecting inputs into a fixed feature space.
result Random Fourier pre-processing accelerates tabular deep learning convergence.
Paper studies IPG method's robustness in noisy distributed linear regression.
problem Distributed linear regression with noise.
method Iteratively Pre-conditioned Gradient-descent (IPG) method.
result IPG method's robustness against noise compared favorably to state-of-the-art algorithms.
New technique explains convergence in ML models with data modifications.
problem Understanding convergence of ML models under data changes.
method Analogue of Fatou's lemma and gamma-convergence.
result Relevance and applications in general ML tasks and domain adaptation.
New method uses off-the-shelf classifiers to improve diffusion generation without extra training.
problem Improving sample quality and controllability in conditional diffusion generation.
method Leveraging off-the-shelf classifiers in a training-free fashion with calibration and pre-conditioning techniques.
result Significant performance improvements (up to 20%) over existing guidance schemes.
We study computational and statistical consequences of problem geometry in stochastic and online optimization. By focusing on constraint set and gradient geometry, we characterize the problem families for which stochastic- and adaptive-gradient methods are (minimax) optimal and, conversely, when nonlinear updates -- su…
The paper uses RL to verify hypotheses, overcoming existing limitations.
problem Verifying hypotheses using reinforcement learning.
method Formulated hypothesis verification as an RL problem, exploiting hypothesis structure.
result RL agents can successfully verify hypotheses, even those not factorizable.
APMSqueeze improves Adam for faster training with less communication.
problem Adam's inefficiency in parallel training due to gradient compression limitations.
method Error-compensated gradient compression for Adam-preconditioned Momentum SGD.
result Significantly reduces communication time per epoch without sacrificing accuracy.
Standard gradient descent methods are susceptible to a range of issues that can impede training, such as high correlations and different scaling in parameter space.These difficulties can be addressed by second-order approaches that apply a pre-conditioning matrix to the gradient to improve convergence. Unfortunately, s…
We simplify diffusion models by defining a design space and improving FID scores.
problem Complexity in diffusion-based generative models.
method Defined a design space, improved sampling and training processes, and pre-conditioned score networks.
result Improved FID scores of 1.79 for CIFAR-10 and 1.97 for unconditional settings.
While stochastic gradient descent (SGD) and variants have been surprisingly successful for training deep nets, several aspects of the optimization dynamics and generalization are still not well understood. In this paper, we present new empirical observations and theoretical results on both the optimization dynamics and…
Combining explicit and implicit regularization improves deep learning performance without needing depth.
problem Improving deep learning performance without increasing model complexity.
method Proposes an explicit penalty to mirror implicit regularization bias in adaptive gradient optimizers.
result Single-layer networks can achieve low-rank approximations with similar performance to deep linear networks.
LQF linearizes deep models for better interpretability.
problem Lack of interpretability in deep neural networks.
method Simple modifications to architecture, loss function, and optimization.
result Comparable performance to non-linear fine-tuning, with interpretability.
The study maps ML quality dimensions to fairness, enhancing the QF4SA framework.
problem Ensuring fairness in ML applications at NSOs to avoid social impacts.
method Employing the QF4SA framework, the study maps quality dimensions to fairness and investigates their interactions.
result Fairness is identified as a new quality dimension in the QF4SA framework.
New algorithm reduces sketching dimension to effective problem size.
problem Solving L2-regularized least-squares problems efficiently.
method Randomized algorithm using Gaussian and SRHT embeddings.
result Preserves convergence guarantees with reduced embedding dimension.
This work improves motion planning for quadcopters by learning and reasoning about controller performance.
problem Improving motion planning for quadcopters with safety margins and execution reliability.
method Introspective learning and reasoning to correct execution bias and improve collision checking.
result Substantial reduction in safety margins for motion actions, leading to safer execution.
We demystify attention patterns in multi-head softmax models for linear data.
problem Understanding the training dynamics and emergent patterns in multi-head softmax attention models.
method Extensive empirical experiments and rigorous theoretical analysis.
result Multi-head softmax attention models approximate a debiased gradient descent predictor, outperforming single-head attention and achieving near-Bayesian optimality.
ScaledGD improves gradient descent for ill-conditioned low-rank matrix estimation.
problem Efficiently solving ill-conditioned low-rank matrix estimation problems.
method Scaled Gradient Descent (ScaledGD) with adaptive pre-conditioners.
result Linear convergence rate independent of condition number, low per-iteration cost.
A new method estimates parameters in heavy-tailed corrupted regression with unknown covariance and heterogeneous noise.
problem Estimating parameters in regression with heavy-tailed errors and unknown covariance.
method Near-optimal computationally tractable estimator based on power method and Multiplicative Weight Update algorithm.
result The estimator achieves the optimal statistical rate and breakdown-point under near-optimal sample size.