Machine Learning (ML) solutions are nowadays distributed, according to the so-called server/worker architecture. One server holds the model parameters while several workers train the model. Clearly, such architecture is prone to various types of component failures, which can be all encompassed within the spectrum of a Byzantine behavior. Several approaches have been proposed recently to tolerate Byzantine workers. Yet all require trusting a central parameter server. We initiate in this paper the study of the "general" Byzantineresilient distributed machine learning problem where no individual component is trusted. In particular, we distribute the parameter server computation on several nodes. We show that this problem can be solved in an asynchronous system, despite the presence of 1 3 Byzantine parameter servers and 1 3 * Equal contribution. Authors are listed alphabetically.
In this paper, we study the strategy-proofness properties of the randomized Condorcet voting system (RCVS). Discovered at several occasions independently, the RCVS is arguably the natural extension of the Condorcet method to cases where a deterministic Condorcet winner does not exists. Indeed, it selects the always-existing and essentially unique Condorcet winner of lotteries over alternatives. Our main result is that, in a certain class of voting systems based on pairwise comparisons of alternatives, the RCVS is the only one to be Condorcet-proof. By Condorcet-proof, we mean that, when a Condorcet winner exists, it must be selected and no voter has incentives to misreport his preferences. We also prove two theorems about group-strategyproofness. On one hand, we prove that there is no group-strategy-proof voting system that always selects existing Condorcet winners. On the other hand, we prove that, when preferences have a one-dimensional structure, the RCVS is group-strategy-proof.
Purpose The moderating role of situational context in the effects of electronic word-of-mouth (eWOM) on online purchase intention through brand image has found sparse empirical support. This study thus aims to examine whether situational context affects the direction and strength of the relationships between aspects of eWOM and brand image that lead to online purchase intention. Design/methodology/approach To extend the existing research, the authors tested the model using a sample of 546 online shoppers during the fourth wave of COVID-19 in Ho Chi Minh City. Specifically, the testing of the direct relationships and the mediating role of brand image occurred using measurement and structural models. The authors then created a moderated mediation model to examine the moderating role of situational context. Furthermore, the authors probed the interactions by identifying changes in the relationships from eWOM to online purchase intention through a brand image at different levels of situational context. Findings Without situational context’s moderating effect, brand image positively partially mediated the influence of either eWOM credibility or quantity on intention. Situational context’s moderating effect then explains why high- versus low-level disease-avoidance customers seek less eWOM credibility and more eWOM quantity to develop brand images and shape their intentions. Originality/value The findings have theoretical implications for understanding the pressure of disease avoidance on customers’ online purchase intentions. Among the practical implications of the research are tactics for profit and non-profit purposes.
Machine learning (ML) solutions are nowadays distributed, according to the so-called server/worker architecture. One server holds the model parameters while several workers train the model. Clearly, such architecture is prone to various types of component failures, which can be all encompassed within the spectrum of a Byzantine behavior. Several approaches have been proposed recently to tolerate Byzantine workers. Yet all require trusting a central parameter server. We initiate in this paper the study of the “general” Byzantine-resilient distributed machine learning problem where no individual component is trusted. In particular, we distribute the parameter server computation on several nodes. We show that this problem can be solved in an asynchronous system, despite the presence of $$\frac{1}{3}$$ 1 3 Byzantine parameter servers (i.e., $$n_{ps} > 3f_{ps}+1$$ n ps > 3 f ps + 1 ) and $$\frac{1}{3}$$ 1 3 Byzantine workers (i.e., $$n_w > 3f_w$$ n w > 3 f w ), which is asymptotically optimal. We present a new algorithm, ByzSGD, which solves the general Byzantine-resilient distributed machine learning problem by relying on three major schemes. The first, scatter/gather, is a communication scheme whose goal is to bound the maximum drift among models on correct servers. The second, distributed median contraction (DMC), leverages the geometric properties of the median in high dimensional spaces to bring parameters within the correct servers back close to each other, ensuring safe and lively learning. The third, Minimum-diameter averaging (MDA), is a statistically-robust gradient aggregation rule whose goal is to tolerate Byzantine workers. MDA requires a loose bound on the variance of non-Byzantine gradient estimates, compared to existing alternatives [e.g., Krum (Blanchard et al., in: Neural information processing systems, pp 118-128, 2017)]. Interestingly, ByzSGD ensures Byzantine resilience without adding communication rounds (on a normal path), compared to vanilla non-Byzantine alternatives. ByzSGD requires, however, a larger number of messages which, we show, can be reduced if we assume synchrony. We implemented ByzSGD on top of both TensorFlow and PyTorch, and we report on our evaluation results. In particular, we show that ByzSGD guarantees convergence with around 32% overhead compared to vanilla SGD. Furthermore, we show that ByzSGD’s throughput overhead is 24–176% in the synchronous case and 28–220% in the asynchronous case.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.