Results 121 to 130 of about 109,902 (144)

Energy‐Efficient Hardware Implementation of Spiking‐Restricted Boltzmann Machines Using Pseudo‐Synaptic Sampling

Advanced Intelligent Systems, EarlyView.
In this article, an energy‐efficient hardware implementation of spiking‐restricted Boltzmann machines using the pseudo‐synaptic sampling (PS2) method is presented. In the PS2 method, superior area and energy efficiency over previous approaches, such as the random walk method, are demonstrated, achieving a 94.94% reduction in power consumption during on‐
Hyunwoo Kim, Suyeon Jang, Uicheol Shin, Masatoshi Ishii, Atsuya Okazaki, Megumi Ito, Akiyo Nomura, Kohji Hosokawa, Sungmin Lee, Matthew BrightSky, Sangbum Kim   +10 more
wiley   +1 more source

Algorithmically Enhanced Wearable Multimodal Emotion Sensor

Advanced Intelligent Systems, EarlyView.
This study presents a fully printed, organic wearable sensor for multimodal emotion sensing by noninvasively monitoring physiological indicators like physiological pulse, breathing patterns, and voice signatures. Using LSTM networks and Q‐learning, the system achieves over 91% accuracy, offering a pathway to understanding complex emotions and enhancing
Anand Babu, Getnet Kassahun, Isabelle Dufour, Dipankar Mandal, Damien Thuau   +4 more
wiley   +1 more source

Multi‐Module Micro/Nanorobots for Biomedical and Environmental Remediation Applications

Advanced Intelligent Systems, EarlyView.
Multi‐module microrobots (MNRs) have overcome the limitations of single‐module systems by integrating components such as propeller, actuator, manipulator, and imaging modalities. They show promise in biomedical applications, such as targeted drug delivery and tissue repair, as well as in environmental remediation, including pollutant removal.
Bairong Zhu, Amar Salehi, Lei Xu, Wei Yuan, Tingting Yu   +4 more
wiley   +1 more source

Bioinspired Tactile Object Identification Leveraging Deep Learning and Soft Body Compliance

Advanced Intelligent Systems, EarlyView.
Herein, it is demonstrated that a soft robotic hand, integrated with low‐resolution tactile sensors, can effectively identify a variety of objects with high accuracy by combining multi‐grasp information. Central to this approach is the development of ROSE‐Net, a specialized neural network designed to harness the data from multiple grasps.
Oliver Shorthose, Luca Scimeca, Alessandro Albini, Perla Maiolino   +3 more
wiley   +1 more source

Knowledge Distillation‐Based Zero‐Shot Learning for Process Fault Diagnosis

Advanced Intelligent Systems, EarlyView.
Process and image data are equivalent with the teacher model pretrained on image data. Knowledge distillation transfers normal condition data to the student model. When an unknown fault occurs, differences between the teacher and student models are quantified via gradients to isolate the fault. Data‐driven deep learning is effective in diagnosing known
Yi Liu, Jiajun Huang, Mingwei Jia
wiley   +1 more source

Fairness in Federated Learning: Trends, Challenges, and Opportunities

Advanced Intelligent Systems, EarlyView.
This survey delves into the intricate issues pertinent to fairness in federated learning , where various biasing factors can skew model performance. By systematically analyzing fairness‐aware strategies, evaluation metrics, and future directions, this work identifies pivotal research gaps in existing approaches and sheds light on both challenges and ...
Noorain Mukhtiar, Adnan Mahmood, Quan Z. Sheng   +2 more
wiley   +1 more source

Deep Learning–Powered Robust Tactile Perception: Bridging Graphene Electronic Skin and Dynamic Decoding

Advanced Intelligent Systems, EarlyView.
This article presents a developed methodology to achieve robust tactile perception with spatially designed electronic skin (e‐skin) utilizing laser‐induced graphene (LIG). To enhance LIG‐based e‐skin, the fabrication of the e‐skin is simplified and advanced decoding algorithms are employed, effectively circumventing several intrinsic issues of ...
Gyeonghwa Heo, Jusouk Yoon, Jeonghwa Jeong, Young Woo Kwon, Suck Won Hong   +4 more
wiley   +1 more source

Assessing Large Multimodal Models for One‐Shot Learning and Interpretability in Biomedical Image Classification

Advanced Intelligent Systems, EarlyView.
Image classification plays a pivotal role in biomedical image analysis. Herein, it is shown that large multimodal models, such as GPT‐4, achieve superior performance in one‐shot learning, generalization, interpretability, and text‐driven image classification. Applications span tissue, cell type, cellular state, and disease classification, outperforming
Wenpin Hou, Qi Liu, Huifang Ma, Yilong Qu, Zhicheng Ji   +4 more
wiley   +1 more source

Magnetic Skin for Voice Disorders with Real‐Time Lip Translation

Advanced Intelligent Systems, EarlyView.
Soüti, a novel wearable technology combining flexible, biocompatible magnetic skin patches with a convolutional neural network to translate lip movements into speech with 92.54% accuracy. This innovative approach enhances therapeutic options for speech impairments and democratizes access to advanced healthcare technologies, particularly benefiting ...
Montserrat Ramirez‐De Angel, Alexander Przybysz, Eckaard le Roux, Nora Alamoudi, Li Zhang, Saptami Suresh Shetty, Khaled N. Salama   +6 more
wiley   +1 more source

Applied Artificial Intelligence in Materials Science and Material Design

Advanced Intelligent Systems, EarlyView.
AI‐driven methods are transforming materials science by accelerating material discovery, design, and analysis, leveraging large datasets to enhance predictive modeling and streamline experimental techniques. This review highlights advancements in AI applications across spectroscopy, microscopy, and molecular design, enabling efficient material ...
Emigdio Chávez‐Angel, Martin Børstad Eriksen, Alejandro Castro‐Alvarez, Jose H. Garcia, Marc Botifoll, Oscar Avalos‐Ovando, Jordi Arbiol, Aitor Mugarza   +7 more
wiley   +1 more source