With the arrival of big data era, opportunities as well challenges arise in both industry and academia. As an important service most web applications, accurate real-time recommendation context is high demand. Traditional recommender systems that analyze update models at regular time intervals cannot satisfy requirements modern calling for systems.

Video recommendation has attracted growing attention in recent years. However, conventional techniques have limitations real-time processing, accuracy or scalability for the large-scale video data. To address deficiencies of current systems, we introduce some new to provide and accurate recommendations users system Tencent Inc.. We develop a scalable online collaborative filtering algorithm based upon matrix factorization, with an adjustable updating strategy considering implicit feedback...

Safety is the most important aspect of an autonomous driving platform. Deep neural networks (DNNs) play increasingly critical role in localization, perception, and control these systems. The object detection classification inference are particular importance to construct a precise picture vehicle's surrounding objects. Graphics Processing Units (GPU) well-suited accelerate such DNN-based applications since they leverage data thread-level parallelism GPU architectures. Understanding...

With the rapid development of high-tech in China, field artificial intelligence has also been promoted. Image recognition is an important topic intelligence, which mainly includes two modules: classification and feature extraction; At same time, as research direction AI, deep learning made progress recent years. It widely used image recognition, speech other fields achieved great success. This paper analyzes application from aspects face remote sensing classification, etc. Its purpose to...

As lithium-iron batteries play a crucial role in the growth of electric vehicles, their disposal is projected to increase, posing significant environmental and health risks. Recovering metals that compose these can help mitigate negative impacts mining address raw material shortages. This research employs hydrometallurgy recover lithium, aluminum, iron, copper from electrode mixture spent lithiumiron batteries. The average metal content found for was approximately 5%, 2%, 18%, 16%,...

Surface quality, geometry confinement, internal integrity, and mechanical properties are of great importance for the applications microstructures fabricated by localized electrochemical deposition. This study shows that copper microcolumn grown resuming anode cathode to a given separation when short-circuit contact occurs exhibits nodular structure with microvoids residing on boundaries. Both voids affect apparent Young's modulus microcolumn. A deposition-detection-withdrawal control method,...

The design of multi-Gbps LDPC decoder has become a hot topic in recent years as the demand transformation towards 5G. An energy efficient 18Gbps based on ASIP with half layer paralleled architecture is proposed. feasibility proven by its demonstrator silicon 28nm CMOS technology, record efficiency 18.4 pJ/decoded bit and area 23.8 Gbps/mm2 for ½ coding rate working at 18.4Gbps. With frequency, voltage scaling multi-core management, supports wide range throughput, from 1.8Gbps to measurement...

The design of multi-Gbit/s low-density parity-check code (LDPC) decoders has become a hot topic in recent years to meet the growing demand transformation towards 4G. An area and energy efficient LDPC decoder engine with fully paralleled layered architecture based on an application-specific instruction set processor (ASIP) using Synopsys IP designer is presented. When ASIP core instantiated for 802.11ad, it achieved throughput up 7 Gbit/s at three iterations latency 95 ns, record efficiency...

IEEE 802.11ay is the amendment to 802.11 standard that enables Wi-Fi devices achieve 100 Gbps using unlicensed mm-Wave (60 GHz) band at comparable ranges today's commercial 60 GHz based on 802.11ad standard. In this paper, we propose a full row-based layered LDPC decoder supporting all coding rates for 802.11ay. Taking property of parity check matrix 802.11ay, combining multiple layers into single layer improves hardware utilization hence increases throughput. The throughput further by...

This paper presents an energy-efficient digital signal processor (DSP) for rectangular-to-polar conversion in 802.11ad polar transmitter working on 60 GHz band. Firstly, system simulations with a complete transmission chain are conducted regard to error vector magnitude and output spectrum, which allows systematically optimize the design requirements DSP block. Secondly, algorithm architecture co-optimization block is explored minimize power consumption. Finally, proposed synthesized using...

A complete Digital Front-End (DFE) processor for 60 GHz polar transmitter is presented. It avoids supply modulating, RF limiters, and AM detection circuits, compared to traditional analog-centric architectures. The front-end consists of i) a poly-phase Cascaded Integrator-Comb (CIC) filter spectrum shaping; ii) parallel Coordinate Rotation DIgital Computer (CORDICs) rectangular-to-polar conversion; iii) Power Amplifier (PA) non-linearities pre-distortion units using Look-Up Tables (LUTs)....

Aggressive voltage and frequency scaling are widely utilized to exploit the design margin introduced by process, environment variations. However, beyond critical or results numerous timing errors, hence unacceptable output quality. In this paper, a computation-skip (CS) scheme is proposed for recursive digital signal processors with fixed cycles per instruction (CPI) correct errors. A CORDIC processor CS still functions when sub-critical frequency. It improves EVM 47.9 dB at its most supply...

Massive MIMO has the potential to bring great spectral and energy efficiency improvements, making it a very promising technology for future wireless systems. Essential achieve gains in practice, is ability realize many antenna paths at low complexity. In this paper, we consider of processing semiconductor edge by allowing voltage over-scaling complete signal failures, focusing on per-antenna digital functionality that dominant DSP The impact resulting hardware errors performance transmission...

Massive MIMO (MaMIMO) is a key technology for 5G wireless communication, enabling large increase in both spectral and energy efficiency at the same time. Before it can be deployed, important to find efficient implementation strategies. Because of many antennas, an essential part decreasing complexity, further improving efficiency, optimization digital signal processing (DSP) per-antenna functions. Assuming orthogonal frequency-division multiplexing (OFDM) based MaMIMO system, this paper...

In this paper we analyze the architecture of a 13 bits 4.096 GHz multistage decimation filter for multi-standards radio receivers. The proposed solution uses shift-and-adder high data rate stages and hardware multiply-accumulator low stages. It also explored benefits using Carry-Save format numbers over binary number. chain is implemented in 45 nm CMOS technology, which exploits advantage all architectures exhibit best area-power trade-off. reduces power by 13.7%, compared with conventional...

Many real-world applications have to face the problem of diversity in data formats and semantics. Currently, how deal with heterogeneous effectively is still a big challenge. With rise knowledge graphs, more are built upon graph-like models, which benefit from flexible schemas convenient support for relationship queries. We propose graph-based unifying system unification, helps (1) transform many other into or conversely, graph formats, (2) integrate based on HAO intelligence, achieves...

Accurate determination of the safe-fault failure rate complex digital designs is an exascale problem. We present a novel measurement methodology and results which could have profound impact on performance availability for GPUs in safety critical systems. extend our analysis with in-the-field verification.

This paper presents a HAO-Interaction question answering system, which exploits knowledge based (KBQA) technology to quickly obtain an answer path for the input question, and then creative text generation mechanism acquire final text. The system also provides visibility of on user interface in order facilitate understanding. Different from other KBQA systems, supports users incorporate organizational graph database while accessing all functionalities. In addition, solution implemented does...