A5100600367- Advanced Battery Materials and Technologies
- Advancements in Battery Materials
- Advanced battery technologies research
- Parallel Computing and Optimization Techniques
- Radiation Effects in Electronics
- Ferroelectric and Negative Capacitance Devices
- Evolutionary Algorithms and Applications
- Advanced Data Storage Technologies
- Simulation Techniques and Applications
- Fuzzy Logic and Control Systems
- Vibration and Dynamic Analysis
- Fuel Cells and Related Materials
- Data Stream Mining Techniques
- Engineering Applied Research
- Electrochemical Analysis and Applications
- Wireless Sensor Networks and IoT
- Technology, Environment, Urban Planning
- Advanced Bandit Algorithms Research
- Electric Motor Design and Analysis
- Coastal and Marine Dynamics
- Earthquake and Tsunami Effects
- Underwater Vehicles and Communication Systems
- Neuroethics, Human Enhancement, Biomedical Innovations
- Bioethics and Human Rights Issues
- Advanced biosensing and bioanalysis techniques
Sungkyunkwan University
2022-2025
Korea Advanced Institute of Science and Technology
2024
La Jolla Alcohol Research
2024
National Disaster Management Research Institute
2024
Samsung (South Korea)
2021
Seoul National University
2016
The University of Melbourne
2014
Hanyang University
2009
Modern transformer-based Large Language Models (LLMs) are constructed with a series of decoder blocks. Each block comprises three key components: (1) QKV generation, (2) multi-head attention, and (3) feed-forward networks. In batched processing, generation networks involve compute-intensive matrix-matrix multiplications (GEMM), while attention requires bandwidth-heavy matrix-vector (GEMV). Machine learning accelerators like TPUs or NPUs proficient in handling GEMM but less efficient for GEMV...
Abstract Oxygen redox in layered oxide cathodes for Na‐ion batteries is considered a promising approach improving the energy density. However, oxygen‐redox‐based suffer from sluggish kinetics and undesirable structural change during charge/discharge, leading to poor electrochemical performances. Herein, introducing vacancies (□) transition metal layers enables enhanced oxygen redox‐based performances P3‐type Mn‐based cathode demonstrated. The can play role of local distortion buffers,...
Abstract The Co‐free Li[Ni 0.75 Mn 0.25 ]O 2 (NM7525) cathode offers the significant advantage of low production costs. However, it requires a high‐voltage charging process (≥4.5 V vs Li/Li⁺) to achieve energy density level high‐Ni x Co y z ( ≥ 0.8) cathodes, which leads severe structural and morphological degradation in secondary particles during prolonged cycling. Herein, is demonstrated that formation stable homogeneous cathode‐electrolyte interface (CEI) can effectively suppress large...
We report a simple, low cost, electrochemical deposition method to pattern gold nanoparticles on flexible, transparent, single-walled carbon nanotube (SWCNT) films, and demonstrate the application of gold-patterned SWCNT films as surface-enhanced Raman spectroscopy substrates biosensing electrodes for non-enzymatic glucose detection.
In this paper, we propose a genetic algorithm for LEGO(R) brick layout problem. The task is to build given 3D object with bricks. A modeled as solution combinatorial optimization problem, through intermediate voxelization, which tries maximize the connectivity and then minimize number of used We attack problem in context search. proposed randomized greedy produces initial solutions, solutions are effectively improved by an evolutionary process. New domain-specific methods well, include...
Portfolio optimization is a control problem whose objective to find the optimal strategy for process of selecting proportions assets that can provide maximum return. Conventional approaches formulate as single Markov decision and apply reinforcement learning methods solutions. However, it well known financial markets involve non-stationary processes, leading violations this assumption in these methods. In work, we reformulate portfolio deal with nature markets. our approach, divide long-term...
Although anionic-redox-based layered oxide materials have attracted great attention as promising cathodes for Na-ion batteries because of their high practical capacities, they suffer from undesirable structural degradation, resulting in the poor electrochemical behavior. Moreover, occurrence stable anionic-redox reaction without use expensive elements such Li, Co, and Ni is considered one most important issues high-energy low-cost batteries. Herein, using first-principles calculation various...
Kim, D.S.; D.-H.; Park, J.-R.; Lee, S.-Y., and H.-Y., 2023. Analysis of tsunami characteristics along Korea's southern coast using a hypothetical scenario. In: J.L.; H.; Min, B.I.; Chang, J.-I.; Cho, G.T.; Yoon, J.-S., J. (eds.), Multidisciplinary Approaches to Coastal Marine Management. Journal Research, Special Issue No. 116, pp. 343-347. Charlotte (North Carolina), ISSN 0749-0208. Large-scale earthquakes are occurring globally, especially in the South Asian crust, which is experiencing...
Modern transformer-based Large Language Models (LLMs) are constructed with a series of decoder blocks.Each block comprises three key components: (1) QKV generation, (2) multi-head attention, and (3) feed-forward networks.In batched processing, generation networks involve compute-intensive matrix-matrix multiplications (GEMM), while attention requires bandwidth-heavy matrix-vector (GEMV).Machine learning accelerators like TPUs or NPUs proficient in handling GEMM but less efficient for GEMV...
As DNNs are widely adopted in various application domains while demanding increasingly higher compute and memory requirements, designing efficient performant NPUs (Neural Processing Units) is becoming more important. However, existing architectural NPU simulators lack support for high-speed simulation, multi-core modeling, multi-tenant scenarios, detailed DRAM/NoC and/or different deep learning frameworks. To address these limitations, this work proposes ONNXim, a fast cycle-level simulator...
Wireless sensor technologies have become popular in recent years as the cost of electronics decreases. In addition, various sensors were incorporated into a smartphone, which caused integration many applications one device. Navigation is typical application smartphone. A common assumption navigation that person's position has to be known exactly so most papers focused on localization process itself rather than customizing it for specific application. this paper we will show complex systems...
We suggest a method of inspecting the latent space stock market data using genetic programming. Given black box patterns and (stock, day) tuples relation matrix is constructed. Applying low-rank factorization technique to induces vector space. By manipulating representations patterns, geometry can be examined. Genetic programming constructs tree representation corresponding an arbitrary representation, allowing us interpret result inspection.