Results 71 to 80 of about 58,776 (261)

Efficient Incremental Learning Using Dynamic Correction Vector

IEEE Access, 2020
One major challenge for modern artificial neural networks (ANNs) is that they typically does not handle incremental learning well. In other words, while learning the new features, the performances of existing features usually deteriorate. This phenomenon
Yun Xiang, Yongbiao Miao, Jingyin Chen, Qi Xuan   +3 more
doaj   +1 more source

Foundational Machine‐Learning Interatomic Potential for Simulating Chemically Complex Ni‐Based Superalloys

Advanced Engineering Materials, EarlyView.
We apply a foundational machine‐learning interatomic potential based on the graph atomic cluster expansion (GRACE) to simulate the commercial Ni‐based single‐crystal superalloy CMSX‐4. Hybrid Monte‐Carlo/molecular dynamics sampling resolves short‐range order in the γ phase and L12 sublattice occupancies in the γ’ phase and connects them to stacking ...
Aditya Vishwakarma, Sarath Menon, Fritz Körmann, Thomas Hammerschmidt, Ralf Drautz   +4 more
wiley   +1 more source

All‐in‐One Analog AI Hardware: On‐Chip Training and Inference with Conductive‐Metal‐Oxide/HfOx ReRAM Devices

Advanced Functional Materials, EarlyView.
An all‐in‐one analog AI accelerator is presented, enabling on‐chip training, weight retention, and long‐term inference acceleration. It leverages a BEOL‐integrated CMO/HfOx ReRAM array with low‐voltage operation (<1.5 V), multi‐bit capability over 32 states, low programming noise (10 nS), and near‐ideal weight transfer.
Donato Francesco Falcone, Victoria Clerico, Wooseok Choi, Tommaso Stecconi, Folkert Horst, Laura Bégon‐Lours, Matteo Galetta, Antonio La Porta, Nikhil Garg, Fabien Alibart, Bert Jan Offrein, Valeria Bragaglia   +11 more
wiley   +1 more source

Hybrid-Based Machine Incremental Learning in K-Nearest Neighbor Heterogeneous Drifting Environment

Applied Sciences
The ability to continuously learn over time by incorporating new information while holding onto previously acquired expertise is known as incremental learning (IL).
Japheth Otieno Ondiek, Kennedy Odhiambo Ogada, Tobias Mwalili   +2 more
doaj   +1 more source

Solution‐Processed Two‐Dimensional Indium Oxide on Sodium‐Embedded Alumina for Reconfigurable Optoelectronic Synaptic Transistors

Advanced Functional Materials, EarlyView.
Wafer‐scale two‐dimensioanl In2Se3 oxidized into InOx on sodium‐embedded beta‐alumina enables multifunctional reconfigurable electronics. Sodium ions accumulate within distinct spatial distribution under drain‐controlle and gate‐controlled operation. Drain‐control operation gives controllability of ultraviolet‐driven optoelectronic synaptic conductance
Jinhong Min, Yonghyun Albert Kwon, Hyunbin Kim, Jihyun Kim, Myeongjin Jung, Jaewon Heo, Gwang Ya Kim, Vlastimil Mazanek, Zdeněk Sofer, In Soo Kim, Jeong Ho Cho, Hyesung Park, Sang Min Won, Joohoon Kang   +13 more
wiley   +1 more source

Robust and Adaptive Incremental Learning for Varying Feature Space

IEEE Access
Real-world multiple or streaming tabular datasets, such as electronic health records from various sources and internet-of-things data generated from different devices, typically exhibit varied feature spaces depending on the datasets. Batch-mode learning
Cheol Ho Kim, Jung-Hoon Lee, Hawon Shin, Ock Kee Baek   +3 more
doaj   +1 more source

Learning Automata Based Incremental Learning Method for Deep Neural Networks

IEEE Access, 2019
Deep learning methods have got fantastic performance on lots of large-scale datasets for machine learning tasks, such as visual recognition and neural language processing. Most of the progress on deep learning in recent years lied on supervised learning,
Haonan Guo, Shilin Wang, Jianxun Fan, Shenghong Li   +3 more
doaj   +1 more source

Integrated Field‐Free SOT Domain‐Wall Synapses and MTJ Stochastic Neurons for Hardware Boltzmann Machines

Advanced Functional Materials, EarlyView.
Field‐free spin‐orbit torque domain‐wall synapses integrated with stochastic MTJ neurons enable compact hardware Boltzmann machines. Leveraging intrinsic stochasticity and multi‐level conductance, the system achieves efficient probabilistic learning with high accuracy, demonstrating a scalable spintronic platform for energy‐efficient edge AI.
Aijaz H. Lone, Meng Tang, Camelia Florica, Bin He, Jingkai Xu, Selma Amara, Georgian Melinte, Xixiang Zhang, Gianluca Setti   +8 more
wiley   +1 more source

Balanced Contrast Class‐Incremental Learning

CAAI Transactions on Intelligence Technology
Continual learning aims to empower a model to learn new tasks continuously while reducing forgetting to retain previously learnt knowledge. In the context of receiving streaming data that are not constrained by the independent and identically distributed
Shiqi Yu, Luojun Lin, Yuanlong Yu
doaj   +1 more source

Spectrally Selective Bipolar Self‐Powered Opto‐Synaptic Responses in Tandem 2D van Der Waals Ferroelectric Heterojunctions

Advanced Functional Materials, EarlyView.
Tandem 2D vdW ferroelectric heterojunctions enable spectrally selective, self‐powered bipolar opto‐synaptic responses for retinal ON/OFF bipolar cell emulation and wavelength‐programmable logic operations. This work establishes vdW ferroelectric heterojunctions as a compelling materials platform for energy‐efficient, adaptive, and human‐like artificial
Rajiv Kumar Pandey, Chang Yong Park, Cheolhwa Jang, Taeho Kang, Sungjoo Lee   +4 more
wiley   +1 more source