Huawei

About: Huawei is a company organization based out in Shenzhen, China. It is known for research contribution in the topics: Terminal (electronics) & Node (networking). 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..

AET vs. AED: Unsupervised Representation Learning by Auto-Encoding Transformations Rather Than Data

Abstract: The success of deep neural networks often relies on a large amount of labeled examples, which can be difficult to obtain in many real scenarios. To address this challenge, unsupervised methods are strongly preferred for training neural networks without using any labeled data. In this paper, we present a novel paradigm of unsupervised representation learning by Auto-Encoding Transformation (AET) in contrast to the conventional Auto-Encoding Data (AED) approach. Given a randomly sampled transformation, AET seeks to predict it merely from the encoded features as accurately as possible at the output end. The idea is the following: as long as the unsupervised features successfully encode the essential information about the visual structures of original and transformed images, the transformation can be well predicted. We will show that this AET paradigm allows us to instantiate a large variety of transformations, from parameterized, to non-parameterized and GAN-induced ones. Our experiments show that AET greatly improves over existing unsupervised approaches, setting new state-of-the-art performances being greatly closer to the upper bounds by their fully supervised counterparts on CIFAR-10, ImageNet and Places datasets.

Multimodal End-to-End Autonomous Driving

Abstract: A crucial component of an autonomous vehicle (AV) is the artificial intelligence (AI) is able to drive towards a desired destination. Today, there are different paradigms addressing the development of AI drivers. On the one hand, we find modular pipelines, which divide the driving task into sub-tasks such as perception and maneuver planning and control. On the other hand, we find end-to-end driving approaches that try to learn a direct mapping from input raw sensor data to vehicle control signals. The later are relatively less studied, but are gaining popularity since they are less demanding in terms of sensor data annotation. This paper focuses on end-to-end autonomous driving. So far, most proposals relying on this paradigm assume RGB images as input sensor data. However, AVs will not be equipped only with cameras, but also with active sensors providing accurate depth information (e.g., LiDARs). Accordingly, this paper analyses whether combining RGB and depth modalities, i.e. using RGBD data, produces better end-to-end AI drivers than relying on a single modality. We consider multimodality based on early, mid and late fusion schemes, both in multisensory and single-sensor (monocular depth estimation) settings. Using the CARLA simulator and conditional imitation learning (CIL), we show how, indeed, early fusion multimodality outperforms single-modality.

MCS Selection for Throughput Improvement in Downlink LTE Systems

Abstract: In this paper, we investigate resource block (RB) assignment and modulation-and-coding scheme (MCS) selection to maximize downlink throughput of long-term evolution (LTE) systems, where all RB's assigned to the same user in any given transmission time interval (TTI) must use the same MCS. We develop several effective MCS selection schemes by using the effective packet-level SINR based on exponential effective SINR mapping (EESM), arithmetic mean, geometric mean, and harmonic mean. From both analysis and simulation results, we show that the system throughput of all the proposed schemes are better than that of the scheme in [7]. Furthermore, the MCS selection scheme using harmonic mean based effective packet-level SINR almost reaches the optimal performance and significantly outperforms the other proposed schemes.

Power and Channel Allocation for Cooperative Relay in Cognitive Radio Networks

Abstract: In this paper, we investigate power and channel allocation for cooperative relay in a three-node cognitive radio network. Different from conventional cooperative relay channels, cognitive radio relay channels can be divided into three categories: direct, dual-hop, and relay channels, which provide three types of parallel end-to-end transmission. In the context, those spectrum bands available at all three nodes may either perform relay diversity transmission or assist the transmission in direct or dual-hop channels. On the other hand, the relay node involves both dual-hop and relay diversity transmission. In this paper, we develop power and channel allocation approaches for cooperative relay in cognitive radio networks that can significantly improve the overall end-to-end throughput. We further develop a low complexity approach that can obtain most of the benefits from power and channel allocation with minor performance loss.

Method and Apparatus for Network Slicing

Abstract: Systems and methods for management of network slices in a communication network such as a 5th generation wireless communication network are provided. Network slicing formats of varying degrees of specificity are defined. An appropriate format may be selected for definition of a network slice. Different segments of a slice can be defined using different formats. Slice scoping, purposing, granularity, and resource usage are described. Slice creation and adaptation, and cloud resource management are also described.