A5114430047- CCD and CMOS Imaging Sensors
- Advanced Neural Network Applications
- Optical Systems and Laser Technology
- Advanced Surface Polishing Techniques
- Advanced Vision and Imaging
- Advanced materials and composites
- Advanced Optical Sensing Technologies
- Network Traffic and Congestion Control
- Advanced Memory and Neural Computing
- Remote Sensing and LiDAR Applications
- Force Microscopy Techniques and Applications
- High-Temperature Coating Behaviors
- High Entropy Alloys Studies
- Catalytic C–H Functionalization Methods
- Superconductivity in MgB2 and Alloys
- VLSI and Analog Circuit Testing
- Diamond and Carbon-based Materials Research
- Intelligent Tutoring Systems and Adaptive Learning
- Handwritten Text Recognition Techniques
- Image and Video Quality Assessment
- Ocular and Laser Science Research
- Advanced Semiconductor Detectors and Materials
- Neural Networks and Applications
- Image Enhancement Techniques
- Brain Tumor Detection and Classification
Tsinghua University
2023-2025
Changchun University of Science and Technology
2017-2025
Tongji University
2025
Tongji Hospital
2025
Chengdu Tool Research Institute (China)
2023-2024
Zhejiang University of Technology
2015-2021
KTH Royal Institute of Technology
2020
Chengdu University of Information Technology
2019
Beijing University of Posts and Telecommunications
2011
Rochester Institute of Technology
1994
This paper presents a 2-to-8-b scalable digital SRAM-based CIM macro that is co-designed with multiply-less neural-network (NN) design methodology and incorporates dynamic-logic-based approximate circuits for vector-vector operations. Digital CIMs enable high throughput reliable matrix-vector multiplications (MVMs); however, face three major challenges to obtain further aggressive gains over conventional architectures: (1) prior exploiting computation suffer from accuracy degradation [1];...
Abstract A mild, external photocatalyst‐ and additive‐free protocol for photo‐induced alkylation/cyclization of unactivated alkenes with halides has been developed. This strategy showed excellent regioselectivity simple operation to synthesize alkyl‐substituted quinazolinones a broad substrate scope. More importantly, chlorinated alkanes were also compatible this transformation.
3D processing plays an important role in many emerging applications such as autonomous driving, visual navigation and virtual reality. Recent research shows that adopting voxel-based sparse convolution (SCONV) a backbone can achieve better performance than point-based network large-scale outdoor scenarios [1]. Moreover, 2D SCONVs are still necessary for Bird's-Eye-View (BEV) neck layers or fusion with image processing. Hardware acceleration is needed multiple key operations, including...
A series of five-element CoCrFeNi-Nbx (x = 0, 1, 3, 5, 7, and 9 wt%) high-entropy alloys were prepared using high-energy ball milling discharge plasma sintering methods. Then, the effects doping with Nb elements on organization properties CoCrFeNi HEAs systematically investigated by tensile testing, hardness examining their micro-morphologies. The results show that addition element, lattice distortion alloy due to large size atoms causes microstructure change from a single-phase FCC...
In this article, a novel flexure-based 3-degree-of-freedom elliptical micro/nano-positioning motion stage was developed to obtain translation motions along the x, y, and z directions with high working bandwidth. The configurations XYZ vertical ensure crosstalk between each axis is small enough. driven by three piezoelectric actuators. corresponding principles stiffness modeling were detailed in article. addition, experimental testing conducted investigate performances of stage. obtained...
To solve the problems of low efficiency and easy error manual bundling, a binding mechanism aviation wire harness was designed. During test, we found that parameter tuning problem PID controller led to its insufficient accuracy stability, which failure bundling. Therefore, use particle swarm optimization tune parameters, but it is not ideal. poor balance between exploration exploitation falling into local optimum, have proposed an improved (GRPSO). GRPSO improves above by introducing...
Large Language Models (LLMs) have demonstrated remarkable performance across a wide range of natural language processing tasks. However, their enormous parameter size and extremely high requirements for compute power pose challenges practical deployment. Recent research has revealed that specific capabilities LLMs, such as numerical reasoning, can be transferred to smaller models through distillation. Some studies explore the potential leveraging LLMs perform table-based reasoning. there...
Fine-grained mixed-precision fully-quantized methods have great potential to accelerate neural network training, but existing exhibit large accuracy loss for more complex models such as diffusion networks. This brief introduces a training accelerator It features novel framework with tensor-type- and noise-strength-aware precision scheduling optimize bit-width allocation. The processing cluster design enables dynamical switching mappings model weights, allows concurrent in 4 different...
Voxel-based point cloud networks composed of multiple kinds sparse convolutions (SCONVs) play an essential role in emerging applications such as autonomous driving and visual navigation. Many researchers have proposed processors for image applications. However, they cannot properly deal with three problems the cloud, including low efficiency random memory access, non-parallel neighbor search area overhead supporting hybrid operators, unbalanced workload among cores. In this work, a 2-D/3-D...
Aligning robot behavior with human preferences is crucial for deploying embodied AI agents in human-centered environments. A promising solution interactive imitation learning from intervention, where a expert observes the policy's execution and provides interventions as feedback. However, existing methods often fail to utilize prior policy efficiently facilitate learning, thus hindering sample efficiency. In this work, we introduce MEReQ (Maximum-Entropy Residual-Q Inverse Reinforcement...
Neural Radiance Fields have achieved success in creating powerful 3D media representations with their exceptional reconstruction capabilities. However, the computational demands of volume rendering pose significant challenges during model training. Existing acceleration techniques often involve redesigning architecture, leading to limitations compatibility across different frameworks. Furthermore, these methods tend overlook substantial memory costs incurred. In response challenges, we...
A Charge Injection Device Technologies 512 X CID array has been tested at liquid nitrogen temperature to assess its suitability for use in low light applications. The CID-38 chip used was installed a SICam dewar/electronics from CIDTEC. For this particular system the noise floor measured be approximately equals 260 electrons rms single readout. Using nondestructive readout option an (root)N, where N is number of readouts, improvement figure can achieved. average dark current 77 degree(s)K...
Arrays of silicon sensors arranged in a CCD focal plane architecture have become the detector choice for astronomical imaging 300 nm to 1 micron region. However other architectures offer attractive additional features such as allowing random access to, and readout of, any subarray on chip, nondestructive interrogation signal level pixel. A 512 x pixel array utilizing Charge Injection that offers advantages has been tested at liquid nitrogen temperature its suitability use astronomy....
The demand for high-quality video has been posing remarkable new challenges on achieving efficiency and fair utilization of network resources. High-quality IPTV VoD high-bandwidth requirements. Another trend is the popularity p2p systems it could severely strain broadband networks. In particular, without ability to explicitly communicate with providers, P2P applications mainly depend inefficient inference network- oblivious peering, leading potential inefficiencies both providers. this...