A5100780888- Advanced Memory and Neural Computing
- CCD and CMOS Imaging Sensors
- Advanced Vision and Imaging
- Advanced Neural Network Applications
- Advanced Sensor and Control Systems
- Fuel Cells and Related Materials
- Image Processing Techniques and Applications
- Advanced battery technologies research
- Corrosion Behavior and Inhibition
- Advanced Image and Video Retrieval Techniques
- Advanced Image Processing Techniques
- Domain Adaptation and Few-Shot Learning
- Power Systems and Technologies
- Image Enhancement Techniques
- Advanced Optical Imaging Technologies
- Machine Learning and Data Classification
- Gaze Tracking and Assistive Technology
- Hydrogen embrittlement and corrosion behaviors in metals
- Electrocatalysts for Energy Conversion
- Visual Attention and Saliency Detection
- Parallel Computing and Optimization Techniques
- COVID-19 diagnosis using AI
- Advanced Data Storage Technologies
- Optical Coherence Tomography Applications
- Advanced Numerical Analysis Techniques
META Health
2022-2025
Guangdong Ocean University
2020-2024
Hasso Plattner Institute
2023-2024
University of Potsdam
2023-2024
McGill University
2024
Meta (United States)
2020-2024
Beijing Normal University
2021-2024
Zhengzhou University
2023
Tianjin University of Commerce
2023
University of Michigan–Ann Arbor
2015-2022
A hybrid catalyst structure can provide abundant active sites and tailored electronic properties, but the major challenge lies in achieving delicate control over its composition architecture to improve catalytic activity toward different electrochemical reactions simultaneously. Herein, we present rational design of a magic with low Pt loading (5.90 wt%), composed CoPt3 CoPt nanoparticles supported on N-doped carbon (CoPt3/CoPt⊂PLNC). Importantly, it shows superior multifunctional alkaline...
Hydrazine-assisted hybrid water electrolysis is an energy-saving approach to produce high-purity hydrogen, whereas the development of pH-universal bifunctional catalysts encounters a grand challenge. Herein, phase-selective synthesis ruthenium phosphide compounds with carbon forming pancake-like particles (denoted as Rux P/C-PAN, x = 1 or 2) presented. The obtained RuP/C-PAN exhibits highest catalytic activity among control samples, delivering ultralow cell voltages 0.03, 0.27, and 0.65 V...
A dual-slope capacitance-to-digital converter for pressure-sensing is presented and demonstrated in a complete microsystem. The design uses base capacitance subtraction with configurable capacitor bank to narrow down input range reduce conversion time. An energy-efficient iterative charge method proposed, employing current mirror that leverages the 3.6 V battery supply available system. We also propose dual-precision comparators comparator power while maintaining high accuracy during slope...
Deep learning has proven to be a powerful tool for wide range of applications, such as speech recognition and object detection, among others. Recently there been increased interest in deep mobile IoT [1] enable intelligence at the edge shield cloud from deluge data by only forwarding meaningful events. This hierarchical thereby enhances radio bandwidth power efficiency trading-off computation communication devices. Since many applications are "always-on" (e.g., voice commands), low is...
The multilayered membranes based on Kevlar nanofibers, cadmium telluride (CdTe) nanocrystals, and phosphoric acid (PA) molecules were prepared with spin coating method. accumulation of nanofibers could support the layered (Kevlar-CdTe-PA)4 membrane. CdTe nanocrystals functioned as bridges to link PA besides improving mechanical property. More captured intermolecular hydrogen bonds while immersed into solution. dominated proton conductivity (Kevlar-CdTe-PA)4/PA membranes....
Simultaneous localization and mapping (SLAM) estimates an agent's trajectory for all six degrees of freedom (6 DoF) constructs a 3D map unknown surrounding. It is fundamental kernel that enables head-mounted augmented/virtual reality devices autonomous navigation micro aerial vehicles. A noticeable recent trend in visual SLAM to apply computationand memory-intensive convolutional neural networks (CNNs) outperform traditional hand-designed feature-based methods [1]. For each video frame,...
We present a millimeter-scale near-field radio system for ultra-low-power (ULP) healthcare sensor nodes. It is specifically designed `syringe implantation' which minimizes invasiveness of implantation. Designing wireless node implanted challenging because: 1) the antenna constrained to diameter syringe needle, significantly constrains link distance through RF. 2) The energy/power are strictly limited by form-factor where thin-film batteries can source only <;10μAh and sustain <;50μA peak...
We present a near-field radio system for millimeter-scale wireless smart sensor node that is implantable through 14-gauge syringe needle. The proposed integrates on chip and magnetic antenna glass substrate within total dimension of 1 × 10 mm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup> . demonstrate energy-efficient active communication between the base station device an RF energy-absorbing tissue. transceiver, digital baseband...
This paper presents a unified 6-D vision processor that enables dense real-time 3-D depth and motion perception at full-high-definition (1920 × 1080, FHD) resolution. The proposed design implements neighbor-guided semi-global matching (NG-SGM) algorithm to unify the stereo optical flow problem reduce computation by 98% compared with original SGM. We introduce new custom-designed, high-bandwidth coalescing crossbar circuit automatically coalesces redundant memory accesses mitigate highly...
We design deep neural networks (DNNs) and corresponding networks' splittings to distribute DNNs' workload camera sensors a centralized aggregator on head mounted devices meet system performance targets in inference accuracy latency under the given hardware resource constraints. To achieve an optimal balance among computation, communication, performance, split-aware architecture search framework, SplitNets, is introduced conduct model designing, splitting, communication reduction...
Marked changes in temperature, pH, dissolved oxygen (DO) content, and nutrient content typically occur marine thermoclines, which are key factors that affect the corrosion of metals. Offshore platforms require metals to be exposed deep-sea environments thus increase their penetration into thermocline. This study investigates galvanic E690 steel a thermocline using simulated (SMT). Specifically, was analyzed wire beam electrode (WBE) technique, linear polarization (LP), morphology, weight...
Extended reality (XR) applications are machine learning (ML)-intensive, featuring deep neural networks (DNNs) with millions of weights, tightly latency-bound (10–20 ms end-to-end), and power-constrained (low tens mW average power). While ML performance efficiency can be achieved by introducing engines within low-power systems-on-chip (SoCs), system-level power for nontrivial DNNs depends strongly on the energy non-volatile memory (NVM) access network weights. This work introduces <italic...
This article presents an energy-efficient deep neural network (DNN) accelerator with non-volatile embedded resistive random access memory (RRAM) for mobile machine learning (ML) applications. DNN implements weight pruning, non-linear quantization, and Huffman encoding to store all weights on RRAM, enabling single-chip processing large models without external memory. A four-core parallel programmable architecture adapts various configurations high utilization. We introduce a customized RRAM...
Augmented Reality/Virtual Reality (AR/VR) glasses are widely foreseen as the next generation computing platform. AR/VR a complex "system of systems" which must satisfy stringent form factor, computing-, power- and thermal- requirements. In this paper, we will show that novel distributed on-sensor compute architecture, coupled with new semiconductor technologies (such dense 3D-IC interconnects Spin-Transfer Torque Magneto Random Access Memory, STT-MRAM) and, most importantly, full...
This paper presents a single-chip, high-performance, and energy-efficient stereo vision depth-estimation processor for micro aerial vehicles (MAVs). The proposed implements the state-of-the-art semi-global matching (SGM) algorithm to deliver full high-definition (HD, 1920 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${\times }$ </tex-math></inline-formula> 1080) stereo-depth outputs with maximum of 38...
Miniature sensor nodes are ideal for monitoring environmental conditions in emerging applications such as oil exploration. One key requirement is embedded non-volatile memory compact and retentive data storage the event that power source exhausted. Non-volatile also allows near-zero standby modes, which particularly challenging to achieve at high temperatures when using SRAM due exponential rise leakage with temperature, rapidly degrades battery life (Fig. 11.2.1). However, traditional NOR...
In realistic metrology, entangled probes are more sensitive to noise, especially for a correlated environment. The precision of parameter estimation with is even lower than that the unentangled ones in this paper, we propose measurement scheme only one auxiliary qubit, which can selectively offset impact environmental noise under situation. We analyse our and find out it approaches Heisenberg limit when prepared proper state. further discuss employing states improve other environment models...
We present a highly flexible micro-robotic vision SoC featuring hybrid Processing Element (PE) for efficient processing of both Convolutional Neural Network (CNN) and non-CNN tasks with 2MB embedded MRAM retentive fully-on-chip weight storage. Fabricated in 22nm, the design achieves 0.22nJ/pix Harris corner detection (a task) 3.5TOPS/W (INT16) CNN, 60% efficiency improvement over state-of-the-art NVM-based NN ASICs.
We propose an ultra-low-power (ULP) image signal processor (ISP) that performs on-the-fly in-processing frame compression/decompression and hierarchical event recognition to exploit the temporal spatial sparsity in sequence. This approach reduces energy consumption spent processing transmitting unimportant data achieve a 16 × imaging system gain intruder detection scenario. The ISP was fabricated 40-nm CMOS consumes only 170 μW at 5 frames/s for neural network-based 192 compressed recording.
Marked changes in temperature, pH, dissolved oxygen (DO) content, and nutrient content typically occur marine thermoclines, which are key factors that affect the corrosion of metals. Offshore platforms have required metals to be exposed deep-sea environments thus increased their penetration into thermocline. This study investigates galvanic E690 steel a thermocline using simulated (SMT). Specifically, was analyzed wire beam electrode (WBE) technique, linear polarization (LP), morphology,...