Results 231 to 240 of about 6,207,310 (278)

Vector‐Based SparseGPT: A Vector‐Based Sparsity Method for Large Language Model

Advanced Intelligent Systems, EarlyView.
Conventional pruning such as SparseGPT produces harmful accumulated errors through step‐by‐step weight editing. V‐SparseGPT prunes and updates target weights simultaneously to curb these errors for N:M structured sparsity. Block‐sparsity and layer‐parallel sparsity greatly boost pruning efficiency.
Junfei He, Mingxi Tan, Junchao Dong, Pengfei Ying   +3 more
wiley   +1 more source

The recognition method of external force damage sources vibration signals based on LSTM-CNN-CatBoost-GSSSA. [PDF]

PLoS One
Chen X, Zhang L, Li X, Gao F.
europepmc   +1 more source

A machine learning approach to risk based asset allocation in portfolio optimization. [PDF]

Sci Rep
Agal S, Raulji K, Odedra ND.
europepmc   +1 more source

Resonance-Anchored Residual Surrogate for Spectral Prediction and Inverse Design of Multiple Terahertz Metamaterial Absorber Topologies. [PDF]

Nanomaterials (Basel)
Li G, Qu W, Huang Y, Wang Y, Yi Q, Deng H.   +5 more
europepmc   +1 more source

Biomimetic Dual-Strategy Adaptive Differential Evolution for Joint Kinematic-Residual Calibration with a Neuro-Physical Hybrid Jacobian. [PDF]

Biomimetics (Basel)
Ma X, Zhao Y, Li Z.
europepmc   +1 more source

Predicting continuous load-settlement trajectories of bored piles: A GA-optimized weighted ensemble framework with explainable AI. [PDF]

PLoS One
Giap LV, Nguyen TL, Pham TA.
europepmc   +1 more source

Temporal speckle enables probabilistic weights for uncertainty-aware photonic AI

Pernice W, Brückerhoff-Plückelmann F, Borras H, Hulyal SU, Meyer L, Ji X, Hu J, Sun J, Klein B, Ebert F, Dijkstra J, McRae L, Schmidt P, Kippenberg T, Fröning H.   +14 more
europepmc   +1 more source

Exploiting Operation Importance for Differentiable Neural Architecture Search [PDF]

IEEE Transactions on Neural Networks and Learning Systems, 2022
Recently, differentiable neural architecture search methods significantly reduce the search cost by constructing a super network and relax the architecture representation by assigning architecture weights to the candidate operations. All the existing methods determine the importance of each operation directly by architecture weights.
Yuan Zhou, Sun Yuan Kung, Xukai Xie
exaly   +6 more sources

SurgeNAS: A Comprehensive Surgery on Hardware-Aware Differentiable Neural Architecture Search

IEEE Transactions on Computers, 2023
This work is partially supported by the Ministry of Education, Singapore, under its Academic Research Fund Tier 2 (MOE2019-T2-1-071) and Tier 1 (MOE2019-T1-001-072), and partially supported by Nanyang Technological University, Singapore, under its NAP (M4082282) and SUG (M4082087).
Weichen Liu, Xiangzhong Luo, Di Liu
exaly   +5 more sources