Article
Computer Science, Artificial Intelligence
Yu-Feng Li, Lan-Zhe Guo, Zhi-Hua Zhou
Summary: This paper investigates safe weakly supervised learning, aims to derive safe predictions by integrating multiple weakly supervised learners. A generic ensemble learning scheme is presented to optimize the worst-case performance gain, bringing multiple advantages to safe weakly supervised learning, demonstrated through extensive experiments on various tasks.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2021)
Article
Computer Science, Artificial Intelligence
Shuo Li, Fang Liu, Zehua Hao, Licheng Jiao, Xu Liu, Yuwei Guo
Summary: Self-supervised representation learning is gaining popularity for its superior performance. Information entropy theory is utilized to propose a simple yet effective self-supervised representation learning method called Minimum Entropy (MinEnt). MinEnt aims to reduce information entropy by optimizing the output of the projector towards its nearest minimum entropy. The core steps of MinEnt include batch dimension normalization, computation of the nearest minimum entropy target, and loss computation for network optimization. Experimental results on widely used datasets show competitive performance of our method in a straightforward manner.
PATTERN RECOGNITION
(2023)
Article
Neurosciences
Celia Ruffino, Dylan Rannaud Monany, Charalambos Papaxanthis, Pauline M. Hilt, Jeremie Gaveau, Florent Lebon
Summary: Practice and motor imagery practice have positive effects on the execution of arm movements, but they differ in their impact on movement smoothness. Practice involves online corrections through sensory feedback integration, while motor imagery practice does not possess this ability.
Article
Environmental Sciences
Hui Li, Zhaodong Niu, Quan Sun, Yabo Li
Summary: Space debris detection is crucial for space missions, but noisy labels can cause problems. To overcome this, a novel label-noise learning method called Co-correcting is proposed, which uses two identical networks to mutually correct the noisy labels and mitigate their effects.
Article
Computer Science, Artificial Intelligence
Zhen-Yu Zhang, Peng Zhao, Yuan Jiang, Zhi-Hua Zhou
Summary: This paper investigates the problem of learning from incomplete and inaccurate supervision and proposes novel approaches to tackle this challenge. The approaches are based on the structure information in limited labeled data and the class-prior information in unlabeled data, effectively alleviating the negative influence of label noise.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2022)
Article
Geochemistry & Geophysics
Dario A. B. Oliveira, Daniil G. Semin, Semen Zaytsev
Summary: Seismic exploration heavily relies on high-quality processing of seismic imaging for interpretation. This process is often prone to error due to human intervention, leading to the need for automation. This article proposes a self-supervised two-step approach to attenuate ground-roll noise in seismic prestack images, which is achieved through the use of convolutional neural networks and conditional generative adversarial networks.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Article
Engineering, Industrial
Pengcheng Xia, Yixiang Huang, Zhiyu Tao, Chengliang Liu, Jie Liu
Summary: This paper proposes a digital twin-enhanced semi-supervised framework for label-scarce motor fault diagnosis. A precise motor digital twin model is established based on multi-physics simulation, and knowledge transfer is performed from the virtual space to the physical space. A novel phase-contrastive current dot pattern (PCCDP) representation is proposed to transform three-phase motor stator current to a gray-scale image for efficient processing. Inter-space sample generation, intra-space sample generation, and clustering-based metric learning methods are introduced to improve fault diagnosis performance.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2023)
Article
Neurosciences
Jeffrey Weiler, Paul L. Gribble, J. Andrew Pruszynski
Summary: The study shows that spinal circuits can efficiently control hand movements during dynamic reaching actions, with the control depending on the hand's location relative to the target.
JOURNAL OF NEUROPHYSIOLOGY
(2021)
Article
Computer Science, Artificial Intelligence
Jing Jiang, Weihong Deng
Summary: This paper aims to improve the performance of in-the-wild Facial Expression Recognition (FER) through semi-supervised learning. The proposed Progressive Teacher (PT) algorithm utilizes reliable FER datasets and large-scale unlabeled expression images for effective training, addressing the lack of data and label noise issues. Experimental results validate the effectiveness of the method, achieving state-of-the-art performance.
IEEE TRANSACTIONS ON AFFECTIVE COMPUTING
(2023)
Review
Neurosciences
James Mathew, Frederic Crevecoeur
Summary: The study reviews computational models of reaching adaptation to force fields in humans, highlighting the crucial role of feedback control in the process. By discussing online adaptation in the feedback control system, the study explains trial-by-trial adaptation and improvements in online motor corrections.
FRONTIERS IN HUMAN NEUROSCIENCE
(2021)
Article
Neurosciences
Benjamin Parrell, Hyosub E. Kim, Assaf Breska, Arohi Saxena, Richard Ivry
Summary: Research shows that cerebellar degeneration affects adaptive responses in reaching and speech production, with no significant differences in compensatory responses. Furthermore, individuals with cerebellar degeneration exhibit impairments in feedforward control, while feedback control remains largely intact.
JOURNAL OF NEUROSCIENCE
(2021)
Article
Computer Science, Artificial Intelligence
Victor R. Barradas, Yasuharu Koike, Nicolas Schweighofer
Summary: Inverse models are essential for human motor learning as they map desired actions to motor commands. The shape of the error surface and the distribution of targets in a task play a crucial role in determining the speed of learning.
Article
Computer Science, Interdisciplinary Applications
Hwa-Seob Song, Byung-Ju Yi, Jong Yun Won, Jaehong Woo
Summary: This paper proposes an autonomous VIR robot system with a deep learning algorithm that drives surgical tools to the target blood vessel location while simultaneously performing surgical tool recognition. It overcomes the limitations of traditional master-slave VIR robot systems.
JOURNAL OF COMPUTATIONAL DESIGN AND ENGINEERING
(2022)
Article
Biochemical Research Methods
Bastien Berret, Adrien Conessa, Nicolas Schweighofer, Etienne Burdet
Summary: The study introduces a new stochastic optimal feedforward-feedback control model that can predict the timing and variability of self-paced arm reaching movements carried out with or without visual feedback. The model considers effort and variance minimization as well as the effects of motor and sensory noise on arm movement planning and execution. By elegantly combining both feedforward and feedback control aspects, the SFFC model is able to address issues where previous models may fail, providing a more comprehensive understanding of human motor control.
PLOS COMPUTATIONAL BIOLOGY
(2021)
Review
Computer Science, Interdisciplinary Applications
Veenu Rani, Syed Tufael Nabi, Munish Kumar, Ajay Mittal, Krishan Kumar
Summary: Machine learning has made significant advances in image processing. Supervised learning relies on labeled data, while unsupervised learning learns from unlabeled data. Self-supervised learning is a type of unsupervised learning that enhances computer vision tasks. This review article provides an in-depth exploration of self-supervised learning and its applications, discussing terms, learning types, and challenges encountered in the process.
ARCHIVES OF COMPUTATIONAL METHODS IN ENGINEERING
(2023)