Study explores strategies for randomized allocation in delayed rewards bandits.
problem Understanding the exploration-exploitation tradeoff in randomized strategies with delayed rewards.
method Examines two strategies: updating exploration sequence at every time point vs. updating only when a new reward is observed.
result The strategy updating only when a new reward is observed leads to strong consistency in allocation for a wider scope of situations.
AB dynamically scales gradients to mitigate asynchronous training delays.
problem Gradient delay in asynchronous training reduces model performance.
method Adaptive Braking (AB) dynamically scales gradients based on alignment.
result AB enables training with up to 32 update steps of delay without accuracy loss.
We study distributed stochastic convex optimization under the delayed gradient model where the server nodes perform parameter updates, while the worker nodes compute stochastic gradients. We discuss, analyze, and experiment with a setup motivated by the behavior of real-world distributed computation networks, where the…
We analyze (stochastic) gradient descent (SGD) with delayed updates on smooth quasi-convex and non-convex functions and derive concise, non-asymptotic, convergence rates. We show that the rate of convergence in all cases consists of two terms: (i) a stochastic term which is not affected by the delay, and (ii) a higher …
We analyze the convergence of gradient-based optimization algorithms that base their updates on delayed stochastic gradient information. The main application of our results is to the development of gradient-based distributed optimization algorithms where a master node performs parameter updates while worker nodes compu…
New asynchronous algorithms improve speed in decentralized optimization networks.
problem Hard convergence analysis for asynchronous decentralized optimization.
method Continuized framework to analyze heterogeneous delays in event-driven updates.
result Achieves asynchronous speedup with convergence rate controlled by eigengap weighted by local delays.
We provide tight finite-time convergence bounds for gradient descent and stochastic gradient descent on quadratic functions, when the gradients are delayed and reflect iterates from τ rounds ago. First, we show that without stochastic noise, delays strongly affect the attainable optimization error: In fact, the error…
Develops accelerated fixed-point methods with delayed oracles for scientific computing.
problem Approximating fixed points of nonexpansive operators.
method Combines Nesterov's acceleration and KM iteration with delayed inexact oracles.
result Establishes improved convergence rates for fixed-point approximation.
Recent years have witnessed the surge of asynchronous parallel (async-parallel) iterative algorithms due to problems involving very large-scale data and a large number of decision variables. Because of asynchrony, the iterates are computed with outdated information, and the age of the outdated information, which we cal…
Paper tackles dueling bandits with delayed feedback, revealing preference bias.
problem Real-world dueling bandit applications often face delays in feedback.
method Introduces biased dueling bandit problem with stochastic delayed feedback, presents two algorithms.
result Two algorithms achieve optimal regret bounds for dueling bandit problems with delay.
In this paper we study the problem of convergence and generalization error bound of stochastic momentum for deep learning from the perspective of regularization. To do so, we first interpret momentum as solving an ℓ2-regularized minimization problem to learn the offsets between arbitrary two successive model para…
We develop parallel and distributed Frank-Wolfe algorithms; the former on shared memory machines with mini-batching, and the latter in a delayed update framework. Whenever possible, we perform computations asynchronously, which helps attain speedups on multicore machines as well as in distributed environments. Moreover…
FedBuff improves federated learning scalability with asynchronous updates.
problem Limited scalability of federated learning with synchronous updates.
method Introduces asynchronous updates (staleness) in federated learning.
result Theoretical analysis shows improved convergence rate with boundedness removed.
HS-FNO models non-Markovian PDEs by learning history and future states.
problem Non-Markovian dynamics where future states depend on past history.
method History-Space Fourier Neural Operator (HS-FNO) for delay and memory-driven PDEs.
result HS-FNO achieves lowest aggregate errors across various PDE families.
NAC-FL optimizes model updates in FL systems by adapting compression to network congestion.
problem Federated Learning systems face congestion and delays in data exchanges.
method NAC-FL dynamically adjusts client compression based on network congestion.
result NAC-FL reduces training time and achieves robust performance improvements.
Improves asynchronous federated learning with queuing dynamics.
problem Asynchronous federated learning with varying node computational speeds.
method Proposes a non-uniform sampling scheme for the central server.
result Significant improvement over current algorithms on image classification.
Large-scale machine learning training, in particular distributed stochastic gradient descent, needs to be robust to inherent system variability such as node straggling and random communication delays. This work considers a distributed training framework where each worker node is allowed to perform local model updates a…
Nonconvex and nonsmooth problems have recently attracted considerable attention in machine learning. However, developing efficient methods for the nonconvex and nonsmooth optimization problems with certain performance guarantee remains a challenge. Proximal coordinate descent (PCD) has been widely used for solving opti…
Online learning algorithms have impressive convergence properties when it comes to risk minimization and convex games on very large problems. However, they are inherently sequential in their design which prevents them from taking advantage of modern multi-core architectures. In this paper we prove that online learning …
Study cooperative bandit learning with imperfect communication, achieving near-optimal performance.
problem Real-world distributed decision-making with imperfect communication.
method Proposed decentralized algorithms for three communication scenarios: stochastic networks, random delays, and adversarially corrupted rewards.
result Achieved competitive performance and near-optimal guarantees on group regret.
Gaussian process improves nowcasting of COVID-19 deaths.
problem Correcting for delays in reporting daily deaths.
method Flexible Gaussian process model with latent variables.
result Gaussian process nowcasts outperform other methods.
Adaptive distributed SGD reduces delay in slow workers.
problem Minimizing delay in distributed SGD with stragglers.
method Adaptive policy for varying k to optimize error-runtime trade-off. result Numerical simulations confirm the effectiveness of the adaptive approach.
Ringmaster LMO accelerates training in distributed systems by asynchronously updating neural networks.
problem Asynchronous training in distributed systems where workers compute gradients at different speeds.
method Introduces an asynchronous LMO-based momentum method for unconstrained stochastic nonconvex optimization.
result Establishes convergence guarantees and time complexity bounds for asynchronous LMO-based updates.
A decentralized routing framework for lunar exploration robots.
problem Routing data in intermittent connectivity lunar networks.
method Graph Attention-based Multi-Agent Reinforcement Learning (GAT-MARL).
result Higher delivery rates, no duplications, fewer packet losses.
In value-based reinforcement learning methods such as deep Q-learning, function approximation errors are known to lead to overestimated value estimates and suboptimal policies. We show that this problem persists in an actor-critic setting and propose novel mechanisms to minimize its effects on both the actor and the cr…
We propose Episodic Backward Update (EBU) - a novel deep reinforcement learning algorithm with a direct value propagation. In contrast to the conventional use of the experience replay with uniform random sampling, our agent samples a whole episode and successively propagates the value of a state to its previous states.…
Fast robust subspace tracking in sparse data-dependent noise with near-optimal delay.
problem Robustly tracking time-varying subspaces in the presence of sparse outliers.
method Introduces a fast mini-batch robust ST solution under mild assumptions.
result Provably correct subspace tracking with near-optimal delay and same time complexity as simple PCA.
A new update rule for deep reinforcement learning reduces learning variance and variance in reference signals.
problem Learning variance and incorrect reference signals in deep reinforcement learning.
method t-soft update method inspired by student-t distribution, which reduces extreme updates and accelerates similar updates.
result The t-soft update method outperforms conventional methods in terms of return and variance in PyBullet robotics simulations.
We suggest a general oracle-based framework that captures different parallel stochastic optimization settings described by a dependency graph, and derive generic lower bounds in terms of this graph. We then use the framework and derive lower bounds for several specific parallel optimization settings, including delayed …
Distributed data-parallel algorithms aim to accelerate the training of deep neural networks by parallelizing the computation of large mini-batch gradient updates across multiple nodes. Approaches that synchronize nodes using exact distributed averaging (e.g., via AllReduce) are sensitive to stragglers and communication…
Simplifies RL training with fewer techniques, reducing bias and instability.
problem Training instabilities and high sample complexity in RL.
method Introduced a simple deterministic policy gradient, used propensity estimation, and delayed policy updates.
result Improved performance and reduced sample complexity through these techniques.
Many distributed machine learning (ML) systems adopt the non-synchronous execution in order to alleviate the network communication bottleneck, resulting in stale parameters that do not reflect the latest updates. Despite much development in large-scale ML, the effects of staleness on learning are inconclusive as it is …
Accelerates policy optimization in RL with optimistic and adaptive updates.
problem Improving policy optimization methods in reinforcement learning.
method Integrates foresight into policy improvement step via optimistic and adaptive updates.
result Designs an optimistic policy gradient algorithm, adaptive via meta-gradient learning.
Develops a stochastic approach to financial market delays.
problem Modeling delays in financial markets with multiple assets.
method Introduces a general stochastic framework for information and order execution delays.
result Delayed markets maintain fundamental asset pricing theorems and no asymptotic free lunch condition.
Paper tackles action delays in reinforcement learning, proposing a delay-aware framework.
problem Action delays degrade reinforcement learning performance in real-world systems.
method Formal definition of delay-aware MDP, transformation into standard MDP with augmented states, delay-aware model-based reinforcement learning framework.
result Proposed framework is more efficient in training and transferable between systems with various delay durations.
New insights explain speedup saturation in distributed learning with large batches and delays.
problem Understanding and optimizing speedup in distributed learning with large batches and delays.
method Theoretical analysis of strongly convex, convex, and non-convex settings, considering data sparsity.
result Identification of a data-dependent parameter explaining speedup saturation in both batch size and gradient staleness.
Existing approaches to resource allocation for nowadays stochastic networks are challenged to meet fast convergence and tolerable delay requirements. The present paper leverages online learning advances to facilitate stochastic resource allocation tasks. By recognizing the central role of Lagrange multipliers, the unde…
A novel fully asynchronous scheme for distributed reinforcement learning over networks.
problem Policy evaluation in distributed reinforcement learning over networks.
method Design of a stochastic average gradient (SAG) based distributed algorithm and push-pull augmented graph approach.
result The proposed algorithm converges at a linear rate of \(\mathcal{O}(c^k)\) with \(c\in(0,1)\) and \(k\) increasing by one per node update.
New algorithm tackles delayed feedback in Lipschitz bandits with sublinear regret.
problem Delayed feedback in Lipschitz bandits.
method Design of algorithms for bounded and unbounded stochastic delays.
result Sublinear regret guarantees for both bounded and unbounded delays.
New algorithms ensure fair selection in combinatorial semi-bandit with unrestricted delays.
problem Fair selection in stochastic combinatorial semi-bandit with delayed feedback.
method Introduced merit-based fairness constraints and new bandit algorithms for reward and fairness.
result Achieved sublinear expected reward and fairness regrets with dependence on delay distribution quantiles.
Banker-OMD improves online learning with delayed feedback.
problem Handling delayed feedback in online learning.
method Generalized Online Mirror Descent (OMD) framework.
result Achieves nearly-optimal performance in three bandit scenarios.
New algorithm handles delayed feedback robustly, reducing regret without knowing delay bounds.
problem Bandits with variably delayed feedback, especially excessive delays.
method Implicit exploration scheme, adaptive skipping, drifted regret control.
result Can tolerate arbitrary excessive delays up to order T, reducing regret.
Paper tackles delays in multi-agent reinforcement learning, improving performance.
problem Challenges in reinforcement learning due to delays in real-world systems.
method Proposes a novel framework for multi-agent reinforcement learning with delays, using Delay-Aware Markov Games and centralized-decentralized training.
result Demonstrates significant improvement in performance with delay-aware multi-agent reinforcement learning.
Study on synchronization in financial markets with time delays.
problem Understanding market dynamics and synchronization in financial systems with time delays.
method Examined a system of coupled non-linear delay-differential equations, linearized for small delays, and analyzed collective dynamics using bifurcation diagrams and numerical solutions.
result Demonstrated that limit cycles can be maintained in coupled N-asset models with appropriate parameterization, leading to market synchronization.
Overlap-Local-SGD improves distributed SGD by overlapping communication and computation.
problem High communication delay and node slowdown in distributed SGD.
method Adding an anchor model to synchronize local updates and pull them towards the anchor model.
result Overlap-Local-SGD speeds up distributed training and mitigates straggler effects.
New algorithm tackles stochastic bandits with varying arm-dependent delays.
problem Applying existing algorithms to stochastic delayed bandit settings is restricted by strong assumptions on delay distributions.
method Proposes a simple UCB-based algorithm called PatientBandits that weakens assumptions on delay distributions.
result Provides bounds on regret and performance lower bounds for the PatientBandits algorithm.
BayTiDe discovers time-delayed differential equations from noisy data.
problem Discovering time-delayed differential equations from data with large delays and noise.
method Bayesian inference with a sparsity-promoting prior.
result BayTiDe accurately identifies time-delayed differential equations with accuracy proportional to data resolution.
TSMB handles time delays in multivariate time series data.
problem Varying time delays in multivariate time series data complicate predictions.
method Time Series Model Bootstrap (TSMB) framework for nonparametric time delay estimation.
result TSMB improves model performance in dynamic data environments.