A5043054935- Advanced Neural Network Applications
- Fault Detection and Control Systems
- Domain Adaptation and Few-Shot Learning
- Advanced Image and Video Retrieval Techniques
- Anomaly Detection Techniques and Applications
- Advanced Image Processing Techniques
- Advanced Vision and Imaging
- Human Pose and Action Recognition
- Advanced Sensor and Control Systems
- Adversarial Robustness in Machine Learning
- Grey System Theory Applications
- Recommender Systems and Techniques
- Robotics and Sensor-Based Localization
- Icing and De-icing Technologies
- Machine Fault Diagnosis Techniques
- Advanced Algorithms and Applications
- Video Surveillance and Tracking Methods
- Image Processing Techniques and Applications
- Machine Learning and Data Classification
- Autonomous Vehicle Technology and Safety
- Topic Modeling
- Advanced Measurement and Detection Methods
- IoT and Edge/Fog Computing
- CCD and CMOS Imaging Sensors
- Hydraulic and Pneumatic Systems
Xi'an Aeronautical University
2016-2025
University of Georgia
2024-2025
William & Mary
2019-2024
China Southern Power Grid (China)
2021-2024
Guizhou University
2013-2024
Beihang University
2024
Beijing Jiaotong University
2023
Williams (United States)
2020-2022
Guizhou Electric Power Design and Research Institute
2019-2022
Institute of Software
2022
With the emergence of a spectrum high-end mobile devices, many applications that formerly required desktop-level computation capability are being transferred to these devices. However, executing Deep Neural Networks (DNNs) inference is still challenging considering high and storage demands, specifically, if real-time performance with accuracy needed. Weight pruning DNNs proposed, but existing schemes represent two extremes in design space: non-structured fine-grained, accurate, not hardware...
Model compression techniques on Deep Neural Network (DNN) have been widely acknowledged as an effective way to achieve acceleration a variety of platforms, and DNN weight pruning is straightforward method. There are currently two mainstreams methods representing extremes regularity: non-structured, fine-grained can high sparsity accuracy, but not hardware friendly; structured, coarse-grained exploits hardware-efficient structures in pruning, suffers from accuracy drop when the rate high. In...
We present a framework for detecting and recognizing human activities outdoor video surveillance applications. Our research makes the following contributions: For activity detection tracking, we improve robustness by providing intelligent control fail-over mechanisms, built on top of low-level motion algorithms such as frame differencing feature correlation. recognition, propose an efficient representation that enables recognition different interaction patterns among group people based...
We propose a new model toward improving the quality of image recommendations in social sharing communities like Pinterest, Flickr, and Instagram. Concretely, we Neural Personalized Ranking (NPR) -- personalized pairwise ranking over implicit feedback datasets that is inspired by Bayesian (BPR) recent advances neural networks. further build an enhanced augmenting basic NPR with multiple contextual preference clues including user tags, geographic features, visual factors. In our experiments...
Deep Neural Networks (DNNs) have emerged as the core enabler of many major applications on mobile devices. To achieve high accuracy, DNN models become increasingly deep with hundreds or even thousands operator layers, leading to memory and computational requirements for inference. Operator fusion (or kernel/layer fusion) is key optimization in state-of-the-art execution frameworks, such TensorFlow, TVM, MNN, that aim improve efficiency However, these frameworks usually adopt approaches based...
The rapid development and wide utilization of object detection techniques have aroused attention on both accuracy speed detectors. However, the current state-of-the-art works are either accuracy-oriented using a large model but leading to high latency or speed-oriented lightweight sacrificing accuracy. In this work, we propose YOLObile framework, real-time mobile devices via compression-compilation co-design. A novel block-punched pruning scheme is proposed for any kernel size. To improve...
Recurrent neural networks (RNNs) based automatic speech recognition has nowadays become promising and important on mobile devices such as smart phones. However, previous RNN compression techniques either suffer from hardware performance overhead due to irregularity or significant accuracy loss the preserved regularity for friendliness. In this work, we propose RTMobile that leverages both a novel block-based pruning approach compiler optimizations accelerate inference devices. Our proposed...
Artificial General Intelligence (AGI), possessing the capacity to comprehend, learn, and execute tasks with human cognitive abilities, engenders significant anticipation intrigue across scientific, commercial, societal arenas. This fascination extends particularly Internet of Things (IoT), a landscape characterized by interconnection countless devices, sensors, systems, collectively gathering sharing data enable intelligent decision-making automation. research embarks on an exploration...
Complementary item recommendation finds products that go well with one another (e.g., a camera and specific lens). While complementary items are ubiquitous, the dimensions by which together can vary both product category, making it difficult to detect at scale. Moreover, in practice, user preferences for be complex combinations of quality evidence complementarity. Hence, we propose new neural recommender Encore jointly learn relationships preferences. Specifically, (i) effectively combines...
Though recent years have witnessed remarkable progress in single image super-resolution (SISR) tasks with the prosperous development of deep neural networks (DNNs), learning methods are confronted computation and memory consumption issues practice, especially for resource-limited platforms such as mobile devices. To overcome challenge facilitate real-time deployment SISR on mobile, we combine architecture search pruning propose an automatic framework that derives sparse (SR) models high...
As deep convolutional neural networks (DNNs) are widely used in various fields of computer vision, leveraging the overfitting ability DNN to achieve video resolution upscaling has become a new trend modern delivery system. By dividing videos into chunks and over-fitting each chunk with super-resolution model, server encodes before transmitting them clients, thus achieving better quality transmission efficiency. However, large number expected ensure good quality, which substantially increases...
The engine reverse thrust control is of great significance to the landing deceleration aircraft. action time inverse overcurrent protection changes with size current, and upper lower levels has a high adaptive ability, which can quickly cut out fault near power side. Compared stage current protection, it better meet dual requirements quick movement selectivity distribution network, when system operation mode changes, its sensitivity will not change greatly. Aiming at electrical...
Fine-tuning helps large language models (LLM) recover degraded information and enhance task performance.Although Low-Rank Adaptation (LoRA) is widely used effective for fine-tuning, we have observed that its scaling factor can limit or even reduce performance as the rank size increases. To address this issue, propose RoRA (Rank-adaptive Reliability Optimization), a simple yet method optimizing LoRA's factor. By replacing $\alpha/r$ with $\alpha/\sqrt{r}$, ensures improved Moreover, enhances...
We propose a lightweight system for (i) semi-automatically discovering and tracking bias themes associated with opposing sides of topic; (ii) identifying strong partisans who drive the online discussion; (iii) inferring opinion "regular" participants. By taking just two hand-picked seeds to characterize topic-space (e.g., "pro-choice" "pro-life") as weak labels, we develop an efficient optimization-based propagation method over social/information network. show how this approach leads 20%...
Recently, a new trend of exploring sparsity for accelerating neural network training has emerged, embracing the paradigm on edge. This paper proposes novel Memory-Economic Sparse Training (MEST) framework targeting accurate and fast execution edge devices. The proposed MEST consists enhancements by Elastic Mutation (EM) Soft Memory Bound (&S) that ensure superior accuracy at high ratios. Different from existing works sparse training, this current work reveals importance schemes performance...
In the radar cross section (RCS) prediction of complex target, intensive computational burden occurs while calculating multiple scattering effect. order to overcome large computing, we present program executing on graphics processing units (GPUs). this paper, analyze properties satellite, which antennas are described as cubes and columns, by employing GPU-based combinational method geometrical optics (GO) physical (PO) together with kd-tree technique. Furthermore, due distinctive treatment,...
Recently, Vision Transformer (ViT) has continuously established new milestones in the computer vision field, while high computation and memory cost makes its propagation industrial production difficult. Pruning, a traditional model compression paradigm for hardware efficiency, been widely applied various DNN structures. Nevertheless, it stays ambiguous on how to perform exclusive pruning ViT structure. Considering three key points: structural characteristics, internal data pattern of ViTs,...
With the increasing demand to efficiently deploy DNNs on mobile edge devices, it becomes much more important reduce unnecessary computation and increase execution speed. Prior methods towards this goal, including model compression network architecture search (NAS), are largely performed independently, do not fully consider compiler-level optimizations which is a must-do for acceleration. In work, we first propose (i) general category of fine-grained structured pruning applicable various DNN...
User-generated item lists are a popular feature of many different platforms. Examples include books on Goodreads, playlists Spotify and YouTube, collections images Pinterest, answers question-answer sites like Zhihu. Recommending is critical for increasing user engagement connecting users to new items, but approaches designed the item-based recommendation, without careful consideration complex relationships between items lists. Hence, in this paper, we propose novel user-generated list...