Huawei

About: Huawei is a company organization based out in Shenzhen, China. It is known for research contribution in the topics: Terminal (electronics) & Signal. The organization has 41417 authors who have published 44698 publications receiving 343496 citations. The organization is also known as: Huawei Technologies & Huawei Technologies Co., Ltd..

AI Benchmark: All About Deep Learning on Smartphones in 2019

Abstract: The performance of mobile AI accelerators has been evolving rapidly in the past two years, nearly doubling with each new generation of SoCs. The current 4th generation of mobile NPUs is already approaching the results of CUDA-compatible Nvidia graphics cards presented not long ago, which together with the increased capabilities of mobile deep learning frameworks makes it possible to run complex and deep AI models on mobile devices. In this paper, we evaluate the performance and compare the results of all chipsets from Qualcomm, HiSilicon, Samsung, MediaTek and Unisoc that are providing hardware acceleration for AI inference. We also discuss the recent changes in the Android ML pipeline and provide an overview of the deployment of deep learning models on mobile devices. All numerical results provided in this paper can be found and are regularly updated on the official project website: this http URL.

egoSlider: Visual Analysis of Egocentric Network Evolution

Abstract: Ego-network, which represents relationships between a specific individual, i.e., the ego, and people connected to it, i.e., alters, is a critical target to study in social network analysis. Evolutionary patterns of ego-networks along time provide huge insights to many domains such as sociology, anthropology, and psychology. However, the analysis of dynamic ego-networks remains challenging due to its complicated time-varying graph structures, for example: alters come and leave, ties grow stronger and fade away, and alter communities merge and split. Most of the existing dynamic graph visualization techniques mainly focus on topological changes of the entire network, which is not adequate for egocentric analytical tasks. In this paper, we present egoSlider, a visual analysis system for exploring and comparing dynamic ego-networks. egoSlider provides a holistic picture of the data through multiple interactively coordinated views, revealing ego-network evolutionary patterns at three different layers: a macroscopic level for summarizing the entire ego-network data, a mesoscopic level for overviewing specific individuals' ego-network evolutions, and a microscopic level for displaying detailed temporal information of egos and their alters. We demonstrate the effectiveness of egoSlider with a usage scenario with the DBLP publication records. Also, a controlled user study indicates that in general egoSlider outperforms a baseline visualization of dynamic networks for completing egocentric analytical tasks.

AutoFIS: Automatic Feature Interaction Selection in Factorization Models for Click-Through Rate Prediction

Abstract: Learning feature interactions is crucial for click-through rate (CTR) prediction in recommender systems. In most existing deep learning models, feature interactions are either manually designed or simply enumerated. However, enumerating all feature interactions brings large memory and computation cost. Even worse, useless interactions may introduce noise and complicate the training process. In this work, we propose a two-stage algorithm called Automatic Feature Interaction Selection (AutoFIS). AutoFIS can automatically identify important feature interactions for factorization models with computational cost just equivalent to training the target model to convergence. In the search stage, instead of searching over a discrete set of candidate feature interactions, we relax the choices to be continuous by introducing the architecture parameters. By implementing a regularized optimizer over the architecture parameters, the model can automatically identify and remove the redundant feature interactions during the training process of the model. In the re-train stage, we keep the architecture parameters serving as an attention unit to further boost the performance. Offline experiments on three large-scale datasets (two public benchmarks, one private) demonstrate that AutoFIS can significantly improve various FM based models. AutoFIS has been deployed onto the training platform of Huawei App Store recommendation service, where a 10-day online A/B test demonstrated that AutoFIS improved the DeepFM model by 20.3% and 20.1% in terms of CTR and CVR respectively.

Distributed Resource Allocation Optimization in 5G Virtualized Networks

Abstract: The concepts of network function virtualization and end-to-end network slicing are the two promising technologies empowering 5G networks for efficient and dynamic network/service deployment and management. In this paper, we propose a resource allocation model for 5G virtualized networks in a heterogeneous cloud infrastructure. In our model, each network slice has a resource demand vector for each of its virtual network functions. We first consider a system of collaborative slices and formulate the resource allocation as a convex optimization problem, maximizing the overall system utility function. We further introduce a distributed solution for the resource allocation problem by forming a resource auction between the slices and the data centers. By using an example, we show how the selfish behavior of non-collaborative slices affects the fairness performance of the system. For a system with non-collaborative slices, we formulate a new resource allocation problem based on the notion of dominant resource fairness and propose a fully distributed scheme for solving the problem. Simulation results are provided to show the validity of the results, evaluate the convergence of the distributed solutions, show protection of collaborative slices against non-collaborative slices and compare the performance of the optimal schemes with the heuristic ones.

System, method and device for realizing multimedia service

Abstract: A realization system, method and device for multimedia service are provided. In the realization system for multimedia service, a service middleware receives multimedia service location information updated by users, multimedia service scheduling policy and device maintenance information of a service control proxy and loads them onto a service location register; the service middleware starts up or stops corresponding service control proxy according to device maintenance information of the service control proxy; the service location register authenticates a user multimedia service control request according to multimedia service location information and determines a service control proxy for the user through authentication according to multimedia service scheduling policy; the user multimedia service control request is forwarded to a determined service control proxy; the determined service control proxy provides multimedia service interactive control with an interactive electronic program guide and multimedia service control with a service server. The control flow of multiple multimedia services is unified.