Researchers have introduced H-ViT, a post-training quantization method that uses power-of-two scaling factors to achieve efficient integer-only vision transformer inference ...
Researchers designed Gsformer, combining convolutional networks with self-attention and sparse attention to accurately predict complex traits across six plant and ...
Researchers developed a hybrid intrusion detection architecture combining spider monkey optimization, convolutional networks, and attentive bidirectional sequence learning to secure ...
A dual-timescale spatiotemporal neural network improves lithium-ion battery state-of-health estimation accuracy by combining intra-cycle signal patterns and inter-cycle degradation trajectories.
A frequency-informed transformer architecture analyzing accelerometer vibration data achieves 99.9 percent leak detection accuracy and identifies specific defect categories within ...
