A5019792492- ZnO doping and properties
- Visual perception and processing mechanisms
- Rough Sets and Fuzzy Logic
- Polymer composites and self-healing
- Ga2O3 and related materials
- Quantum Dots Synthesis And Properties
- Synthesis and Properties of Aromatic Compounds
- Copper-based nanomaterials and applications
- Scientific Computing and Data Management
- Multi-Criteria Decision Making
- Axial and Atropisomeric Chirality Synthesis
- Caching and Content Delivery
- Toxic Organic Pollutants Impact
- Cognitive and developmental aspects of mathematical skills
- Advanced Polymer Synthesis and Characterization
- Data Mining Algorithms and Applications
- Advanced Computational Techniques and Applications
- Advanced Neural Network Applications
- Behavioral and Psychological Studies
- Catalytic Processes in Materials Science
- Parallel Computing and Optimization Techniques
- Silicone and Siloxane Chemistry
- GaN-based semiconductor devices and materials
- Advanced Image and Video Retrieval Techniques
- Tensor decomposition and applications
McGill University
2022-2024
Cisco College
2023
Ningbo University
2020
National Chiayi University
2013-2014
Henan Normal University
2014
National Sun Yat-sen University
2010
The surface quality of ZnO nanorod after postannealing was inspected by second harmonic generation (SHG) with the assistance photoluminescence (PL) spectra, and X-ray photoelectron (XPS) spectroscopy. SHG is sensitive to structure, which involves elimination defects restructuring. Oxygen (O)-deficient were generated during growth nanorods. PL XPS analyses indicate that on nanorods reduced at annealing temperatures above 600 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML"...
Room-temperature ferromagnetism (RTFM) was revealed in Fe-doped ZnO vertically aligned nanorod arrays grown by hydrothermal synthesis method. The evolution of RTFM properties and quality were studied different growth temperature. At the temperature 80 °C, nanorods well-formed, Fe readily substituted for Zn arrays. weak room determined magnetization measurements. morphology examined many structure composition analysis tools. atoms found to be incorporated into lattice without any...
We present Mirror, an open-source platform for data exploration and analysis powered by large language models. Mirror offers intuitive natural interface querying databases, automatically generates executable SQL commands to retrieve relevant summarize it in language. In addition, users can preview manually edit the generated ensure accuracy of their queries. also visualizations facilitate understanding data. Designed with flexibility human input mind, is suitable both experienced analysts...
Huge embedding tables in modern Deep Learning Recommender Models (DLRM) require prohibitively large memory during training and inference. Aiming to reduce the footprint of training, this paper proposes FIne-grained In-Training Embedding Dimension optimization (FIITED). Given observation that vectors are not equally important, FIITED adjusts dimension each individual vector continuously assigning longer dimensions more important embeddings while adapting dynamic changes data. A novel storage...
Points of Subjective Equality (PSE) are commonly measured using staircase or constant stimuli methods. However, the method is highly dependent on step size, and time-consuming. Thus, we wanted to develop an efficient quick estimate both PSE slope psychometric function. We developed a random-step algorithm in which one-up-one-down rule followed but with random size pre-defined range test levels. Each stimulus would be chosen depending previous response subject. If subject responded "up", any...
We report on the effects of post-annealing ZnO nanorods array using optical second harmonic generation (SHG). The high-quality and nearly vertically aligned nanorod arrays were deposited ZnO-seeded glass substrates by aqueous solution method. SHG result indicated transformation oxygen deficiency surface after treatment process. chemical composition was investigated X-ray photoelectron spectroscopy. corresponding structural properties scanning electron microscopy diffraction.
Deep neural networks (DNNs) are of critical use in different domains. To accelerate DNN computation, tensor compilers proposed to generate efficient code on domain-specific accelerators. Existing mainly focus optimizing computation efficiency. However, memory access is becoming a key performance bottleneck because the computational accelerators increasing much faster than performance. The lack direct description and data dependence current compilers' intermediate representation (IR) brings...
Stereopsis allows us to judge whether objects are “in front” or “behind” each other. However, it is still not clear this ability position in depth harnessing the same mechanism stimulus “popping in” (crossed disparity) out” (uncrossed disparity). We used a paradigm called "2-by-2 forced-choice paradigm". This involved two intervals of presentations - target and noise. Subjects would indicate which interval appeared was crossed uncrossed. So, we could record when subject sees (detection)...
Points of Subjective Equality (PSEs) are usually measured with constant stimuli (CS) methods. However, it’s found time-consuming and inefficient in clinic settings such as our recent studies investigating the Pulfrich phenomenon (Reynaud, A. & Hess, R. F. (2019), An unexpected spontaneous motion-in-depth amblyopia, Vision 3(4). 54). Thus, we wanted to develop a more efficient method estimate PSE its variability. We have developed an adaptive (AD) which levels chosen on pre-defined scale for...