Session-based recommendation focuses on the prediction of user actions based anonymous sessions and is a necessary method in lack historical data. However, none existing session-based methods explicitly takes long-tail into consideration, which plays an important role improving diversity producing serendipity. As distribution items with prevalent scenarios (e.g., e-commerce, music, TV program recommendations), more attention should be put recommendation. In this paper, we propose novel...

This paper explores meta-learning in sequential recommendation to alleviate the item cold-start problem. Sequential aims capture user's dynamic preferences based on historical behavior sequences and acts as a key component of most online scenarios. However, previous methods have trouble recommending items, which are prevalent those As there is generally no side information task, could not be applied when only user-item interactions available. Thus, we propose Meta-learning-based Cold-Start...

For the virtual synchronous generator, voltage-controlled harmonic suppression method will seriously deteriorate magnitude of inrush fault current in moment grid symmetrical fault, because effects and are mutually constrained from perspective its output impedance. this issue, a coordinated is proposed to achieve both good results normal limitation grid, which mainly includes passive branch, an active branch improved generator (VSG) control. The installed between PCC introduce additional...

The task of session-based recommendation is to predict user actions based on anonymous sessions. Recent research mainly models the target session as a sequence or graph capture item transitions within it, ignoring complex between items in different sessions that have been generated by other users. These include potential collaborative information and reflect similar behavior patterns, which we assume may help with for session. In this paper, propose novel method, namely Dual-channel Graph...

The embedding-based representation learning is commonly used in deep recommendation models to map the raw sparse features dense vectors. traditional embedding manner that assigns a uniform size all has two issues. First, numerous inevitably lead gigantic table causes high memory usage cost. Second, it likely cause over-fitting problem for those do not require too large capacity. Existing works try address always significant drop performance or suffers from limitation of unaffordable training...

Wide area monitoring systems (WAMS) provide effective support for online identification of low‐frequency oscillations in power systems. The WAMS signal is sensitive to the surrounding environment, and it contains Gaussian white noise. noise will produce coloured through filter, which bring some errors mode identification. In order solve above problem, this study proposes a method identify oscillation modes single‐channel measurement system based on combination fourth‐order mixed mean...

Due to the train load stochastic jumps, negative sequence currents (NSCs) are unpredictably brought into power grid, which may attack system and damage vital devices. To enhance compensation capacities improve quality, a hybrid scheme is studied in consideration of composed railway conditioner, thyristor switched capacitor, reactor. First, discrete jump model its principle V/v traction analyzed design compensators capacities. Then, based on attack-defense game theory, two-level optimized...

Predicting user actions based on anonymous sessions is a challenge to general recommendation systems because the lack of profiles heavily limits data-driven models. Recently, session-based methods have achieved remarkable results in dealing with this task. However, upper bound performance can still be boosted through innovative exploration limited data. In paper, we propose novel method, namely Intra-and Inter-session Interaction-aware Graph-enhanced Network, take inter-session item-level...

Siyi Liu, Sihao Chen, Xander Uyttendaele, Dan Roth. Proceedings of the 2021 Conference North American Chapter Association for Computational Linguistics: Human Language Technologies. 2021.

Understanding and characterizing the discus- sions around key events in news streams is important for analyzing political discourse. In this work, we study problem of identification such articles associated with those from streams. We propose a generic framework stream clustering that analyzes temporal trend to automatically extract underlying draw significant media attention. characterize by generating event summaries, based on which form document clusters an unsupervised fashion. evaluate...

Navigation is a primary interaction in virtual reality. Previous research has explored different forms of artificial locomotion techniques for navigation, including hand gestures and body motions. However, few studies have investigated force-based foot as technique. We present three (Foot Fly, Foot Step Teleportation) navigation environment, relying on surface electromyography sensors readings from leg muscles. A pilot study comparing our with controller-based indicates that can provide fun...

Mirroring is the behavior in which one person subconsciously imitates gesture, speech pattern, or attitude of another. In conversations, mirroring often signals speakers enjoyment and engagement their communication. chatbots, methods have been proposed to add personas chatbots train them speak shift dialogue style that personas. However, they require a large dataset consisting dialogues target personalities train. this work, we explore method can learn mirror speaking styles incrementally....

Wireless sensor network (WSN) with several sensors is used to group measurements of certain physical quantities or environmental conditions, including sound, temperature, pressure, vibration, motion, pollution, at various locations and ranges. In WSNs, protocols address the issue routing. Sensor nodes in WSN usually have limited energy resources storage capacities. Therefore, usage very important. This paper analyzes compares quality service (QoS) performances three important routing mobile...

Aiming at the problem that 10kV distribution network branch line ungrounded disconnection fault is difficult to be detected, using multi-source data from network, method of diagnosis through mining proposed. Firstly, based on existing sources, a feature set with both electrical features and non- established. Then, Principal Component Analysis algorithm used for reduction. Finally, FP-Growth algorithm, association rules are mined determine whether has happened. The case analysis historical...

The task of session-based recommendation is to predict user actions based on anonymous sessions. Recent research mainly models the target session as a sequence or graph capture item transitions within it, ignoring complex between items in different sessions that have been generated by other users. These include potential collaborative information and reflect similar behavior patterns, which we assume may help with for session. In this paper, propose novel method, namely Dual-channel Graph...

As a very important network equipment in the information environment, smart phones provide many convenient conditions for people's work and study, which not only makes Internet faster, but also meets needs of daily life through mobile phone software. For development Android software, Java language is generally used to enhance security reliability system. In order improve efficiency software development, this paper discusses practical use work. Firstly, characteristics Software Application...

Recently, diffusion-based generative models have demonstrated remarkable performance in speech enhancement tasks. However, these methods still encounter challenges, including the lack of structural information and poor low Signal-to-Noise Ratio (SNR) scenarios. To overcome we propose Schr\"oodinger Bridge-based Speech Enhancement (SBSE) method, which learns diffusion processes directly between noisy input clean distribution, unlike conventional systems that learn data to Gaussian...