No abstract

Although dispersing one single task to distributed learning nodes has been intensively studied by the previous research, multi-task learning on distributed networks is still an area that has not been fully exploited, especially under decentralized settings. The challenge lies in the fact that different tasks may have different optimal learning weights while communication through the distributed network forces all tasks to converge to an unique classifier. In this paper, we present a novel algorithm to overcome this challenge and enable learning multiple tasks simultaneously on a decentralized distributed network. Specifically, the learning framework can be separated into two phases: (i) multi-task information is shared within each node on the first phase; (ii) communication between nodes then leads the whole network to converge to a common minimizer. Theoretical analysis indicates that our algorithm achieves

Mobile crowdsensing has been intensively explored recently due to its flexible and pervasive sensing ability. Although many crowdsensing platforms have been built for various applications, the general issue of how to manage such systems intelligently remains largely open. While recent investigations mostly focus on incentivizing crowdsensing, the robustness of crowdsensing toward uncontrollable sensing quality, another important issue, has been widely neglected. Due to the non-professional personnel and devices, the quality of crowdsensing data cannot be fully guaranteed, hence the revenue gained from mobile crowdsensing is generally uncertain. Moreover, the need for compensating the sensing costs under a limited budget has exacerbated the situation: one does not enjoy an infinite horizon to learn the sensing ability of the crowd and hence to make decisions based on sufficient statistics. In this paper, we present a novel framework, Budget LImited robuSt crowdSensing (BLISS), to handle this problem through an online learning approach. Our approach aims to minimize the difference on average sense (a.k.a. regret) between the achieved total sensing revenue and the (unknown) optimal one, and we show that our BLISS sensing policies achieve logarithmic regret bounds and Hannan-consistency. Finally, we use extensive simulations to demonstrate the effectiveness of BLISS. Index Terms-Crowdsourcing, machine learning algorithms .1063-6692

With the renaissance of deep learning, the side-channel community also notices the potential of this technology, which is highly related to the profiling attacks in the side-channel context. Many papers have recently investigated the abilities of deep learning in profiling traces. Some of them also aim at the countermeasures (e.g., masking) simultaneously. Nevertheless, so far, all of these papers work with an (implicit) assumption that the number of time samples in raw traces can be reduced before the profiling, i.e., the position of points of interest (PoIs) can be manually located. This is arguably the most challenging part of a practical black-box analysis targeting an implementation protected by masking. Therefore, we argue that to fully utilize the potential of deep learning and get rid of any manual intervention, the end-to-end profiling directly mapping raw traces to target intermediate values is demanded.In this paper, we propose a neural network architecture that consists of encoders, attention mechanisms and a classifier, to conduct the end-to-end profiling. The networks built by our architecture could directly classify the traces that contain a large number of time samples (i.e., raw traces without manual feature extraction) while whose underlying implementation is protected by masking. We validate our networks on several public datasets, i.e., DPA contest v4 and ASCAD, where over 100,000 time samples are directly used in profiling. To our best knowledge, we are the first that successfully carry out end-to-end profiling attacks. The results on the datasets indicate that our networks could get rid of the tricky manual feature extraction. Moreover, our networks perform even systematically better (w.r.t. the number of traces in attacks) than those trained on the reduced traces. These validations imply our approach is not only a first but also a concrete step towards end-to-end profiling attacks in the side-channel context.

This paper presents a control strategy to improve the output power for a single-cylinder two-stroke free-piston linear generator (FPLG). The comprehensive simulation model of this FPLG is established and the operation principle is introduced. The factors that affect the output power are analyzed theoretically. The characteristics of the piston motion are studied. Considering the different features of the piston motion respectively in acceleration and deceleration phases, a ladder-like electromagnetic force control strategy is proposed. According to the status of the linear electric machine, the reference profile of the electromagnetic force is divided into four ladder-like stages during one motion cycle. The piston motions, especially the dead center errors, are controlled by regulating the profile of the electromagnetic force. The feasibility and advantage of the proposed control strategy are verified through comparison analyses with two conventional control strategies via MatLab/Simulink. The results state that the proposed control strategy can improve the output power by around 7-10% with the same fuel cycle mass.