Article
Physics, Fluids & Plasmas
Tomas Barta, Lubomir Kostal
Summary: Strong inhibitory input in balanced neural networks increases synaptic current fluctuations and decreases membrane potential fluctuations, enhancing spike-firing regularity in models with dynamic firing thresholds. This highlights the importance of biologically plausible inputs and specific adaptation mechanisms in neuronal modeling.
Article
Physics, Applied
Tianshi Gao, Bin Deng, Jixuan Wang, Jiang Wang, Guosheng Yi
Summary: This study investigates the propagation of spiking regularity in a feedforward network with recurrent connections, finding that an appropriate excitation-inhibition ratio maximizes regularity in deeper layers. Additionally, collective temporal regularity in deeper layers exhibits resonance behavior with respect to both synaptic connection probability and synaptic weight.
INTERNATIONAL JOURNAL OF MODERN PHYSICS B
(2021)
Article
Engineering, Mechanical
Halgurd Taher, Daniele Avitabile, Mathieu Desroches
Summary: We report a detailed analysis on the emergence of bursting in a recently developed neural mass model that includes short-term synaptic plasticity. The study reveals the importance of synaptic dynamics in bursting activity and the complex process of bursting initiation.
NONLINEAR DYNAMICS
(2022)
Article
Biology
Lu Luo, Xiongfei Wang, Junshi Lu, Guanpeng Chen, Guoming Luan, Wu Li, Qian Wang, Fang Fang
Summary: This study investigates the receptive fields (RFs) of neurons in the human visual cortex using intracranial local field potentials (LFPs) and spiking activity. The results show that the RF sizes and temporal profiles measured from low-gamma activity (LGA) and high-gamma activity (HGA) closely match those measured from spiking activity, suggesting the important role of LGA and HGA in early visual information processing.
SCIENCE CHINA-LIFE SCIENCES
(2023)
Article
Biochemistry & Molecular Biology
Paolo Papale, Feng Wang, A. Tyler Morgan, Xing Chen, Amparo Gilhuis, Lucy S. Petro, Lars Muckli, Pieter R. Roelfsema, Matthew W. Self
Summary: This study reveals that contextual information has a rapid and selective impact on neuronal activity in the V1 region of the brain, it can be used to decode scenes and can be predicted from feedforward input. Additionally, the structure of V1 representations measured with electrophysiology in monkeys correlates strongly with the representations measured with fMRI in humans.
Article
Neurosciences
Salil Bhat, Michael Luhrs, Rainer Goebel, Mario Senden
Summary: Population receptive field (pRF) mapping is a popular tool in computational neuroimaging for studying receptive field properties and topography, with potential applications in brain-computer interface (BCI) communication systems. A novel and fast pRF mapping procedure based on hashed-Gaussian encoding significantly reduces computational resources, facilitating real-time applications.
Article
Multidisciplinary Sciences
Peng-fei Meng, Shuang-cheng Jia, Qian Li
Summary: In this paper, an improved CRNN network CRNN-RES is proposed to address the shortcomings of CRNN in handling wide characters and long dense small characters, which reduces the number of network parameters and improves accuracy. Comparing with the original CRNN network, CRNN-RES achieved higher recognition accuracy on public datasets IC03, IC13, IIIT5k, and SVT.
SCIENTIFIC REPORTS
(2021)
Article
Computer Science, Information Systems
Deming Zhou, Weitong Chen, Kongyang Chen, Bing Mi
Summary: In spiking neural networks, security threats such as adversarial samples and poisoned data can negatively affect model performance. To address this issue, a novel model strengthening method is proposed to efficiently remove malicious data from a trained model.
Article
Multidisciplinary Sciences
Feng Ni, Junnian Wang, Jialin Tang, Wenjun Yu, Ruihan Xu
Summary: In this paper, an improved lightweight convolutional neural network is constructed based on the feature fusion network for side-channel analysis. The experimental results show that the new network has faster convergence, better robustness, and higher accuracy. Compared with traditional neural network methods, it has higher heat value and more concentration in the key interval.
Article
Computer Science, Artificial Intelligence
Yuandong Ma, Qing Song, Hezheng Lin, Chun Liu, Mengjie Hu, Xiaotong Zhu
Summary: This paper proposes a fast and robust feature registration algorithm to address the problem of registration in low-texture cases. The algorithm utilizes a two-step learning framework, adaptive heterogeneous filters, registration feature refiners, and a Warp Extractor to achieve feature fusion and improve the speed of model inference. Experimental results demonstrate that the proposed method outperforms existing state-of-the-art methods and has faster inference speed under different sensor modalities.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Tianshi Gao, Bin Deng, Jixuan Wang, Jiang Wang, Guosheng Yi
Summary: In this study, two-compartment biophysical models were used to construct multilayered feedforward neural networks to investigate how dendritic passive properties affect the propagation of slowly-varying inputs. The results showed that there is an optimal coupling conductance between dendritic and somatic compartments to maximize the fidelity of the initial spiking activity. Increasing the dendritic area enhances the initial firing rate of neurons and promotes the transmission of signals in the neural networks. Changes in the area proportion occupied by somatic compartment and coupling conductance affect the signal propagation ability of the neural networks by adjusting the input-output transform of a single neuron.
Article
Computer Science, Artificial Intelligence
Hayat Yedjour, Boudjelal Meftah, Dounia Yedjour, Olivier Lezoray
Summary: This paper presents a biologically inspired spiking neural network model for detecting motion in image sequences based on MT cell responses. The experimental results demonstrate the network's ability to segregate multiple moving objects and reproduce MT cells' responses. Compared to state-of-the-art methods for boundary detection, the proposed network model provides the best results on the YouTube Motion Boundaries dataset.
NEURAL COMPUTING & APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Xiujian Hu, Guanglei Sheng, Piao Shi, Yuanyuan Ding
Summary: The performance of a convolutional neural network (CNN) model depends on factors such as depth, width, network structure, receptive field size, and feature map scaling. This article analyzes the key factors influencing network performance and proposes strategies for constructing efficient convolutional networks. It introduces a novel architecture called TbsNet, which minimizes computation costs and feature redundancy using lightweight operators.
PEERJ COMPUTER SCIENCE
(2023)
Article
Computer Science, Artificial Intelligence
Sebastian Billaudelle, Benjamin Cramer, Mihai A. Petrovici, Korbinian Schreiber, David Kappel, Johannes Schemmel, Karlheinz Meier
Summary: Neuromorphic devices offer an accelerated and scalable alternative to neural network simulations in computational neuroscience and machine learning, with their neural connectivity and synaptic capacity depending on specific design choices. A strategy has been proposed to achieve structural plasticity by constantly rewiring presynaptic and postsynaptic partners, optimizing resource allocation while maintaining constant neuronal fan-in and sparse connectome. This approach has been implemented on the BrainScaleS-2 system and shown to optimize network topology based on training data nature, demonstrating overall computational efficiency.
Article
Computer Science, Information Systems
Dezhong Xu, Lifang Wu, Yonghao He, Qing Zhao, Meng Jian, Junchi Yan, Liang Zhao
Summary: The paper proposes a Light and Fast Face Detector with an Ommateum Structure (OS-LFFD), which uses a 4-branch network to cover the target range of face sizes and reduce the number of model parameters, achieving a balance between accuracy and running speed on different hardware platforms.
MULTIMEDIA TOOLS AND APPLICATIONS
(2021)
Article
Neurosciences
Anthony Park, Ying Li, Radi Masri, Asaf Keller
JOURNAL OF NEUROPHYSIOLOGY
(2017)
Article
Neurosciences
Alberto Castro, Charles Raver, Ying Li, Olivia Uddin, David Rubin, Yadong Ji, Radi Masri, Asaf Keller
JOURNAL OF NEUROSCIENCE
(2017)
Article
Neurosciences
Wendy A. Friedman, H. Philip Zeigler, Asaf Keller
JOURNAL OF NEUROPHYSIOLOGY
(2012)
Article
Neurosciences
Anthony Park, Kathleen Hoffman, Asaf Keller
JOURNAL OF NEUROPHYSIOLOGY
(2014)
Article
Neurosciences
Jessica L. Whitt, Radi Masri, Nisha S. Pulimood, Asaf Keller
JOURNAL OF NEUROSCIENCE
(2013)
Article
Neurosciences
Masamichi Okubo, Alberto Castro, Wei Guo, Shiping Zou, Ke Ren, Feng Wei, Asaf Keller, Ronald Dubner
JOURNAL OF NEUROSCIENCE
(2013)
Article
Neurosciences
Sylvina M. Raver, Asaf Keller
Article
Neurosciences
Sylvina M. Raver, Sarah P. Haughwout, Asaf Keller
NEUROPSYCHOPHARMACOLOGY
(2013)
Article
Clinical Neurology
Junfang Wu, Charles Raver, Chunshu Piao, Asaf Keller, Alan I. Faden
Article
Neurosciences
Qizong Yang, Chia-Chien Chen, Raddy L. Ramos, Elizabeth Katz, Asaf Keller, Joshua C. Brumberg
SOMATOSENSORY AND MOTOR RESEARCH
(2014)
Article
Neurosciences
Charles Raver, Olivia Uddin, Yadong Ji, Ying Li, Nathan Cramer, Carleigh Jenne, Marisela Morales, Radi Masri, Asaf Keller
JOURNAL OF NEUROSCIENCE
(2020)
Article
Neurosciences
Jason B. Alipio, Catherine Haga, Megan E. Fox, Keiko Arakawa, Rakshita Balaji, Nathan Cramer, Mary Kay Lobo, Asaf Keller
Summary: The study reveals that perinatal fentanyl exposure has lasting consequences on sensory processing and function in mice, impacting their somatosensory function and behavior extending at least to adolescence. This exposure leads to abnormal changes in synaptic excitation, morphological structure, and mRNA expression in the brains of the exposed mice.
JOURNAL OF NEUROSCIENCE
(2021)
Article
Neurosciences
Jason Bondoc Alipio, Lace Marie Riggs, Madeline Plank, Asaf Keller
Summary: The lasting effects of perinatal fentanyl exposure on behavior and synapses in mice can be mitigated by environmental enrichment.
JOURNAL OF NEUROSCIENCE
(2022)
Article
Neurosciences
Yadong Ji, Chimdiya Onwukwe, Jesse Smith, Hanna Laub, Luca Posa, Asaf Keller, Radi Masri, Nathan Cramer
Summary: Noxious stimuli activate catecholaminergic input and increase the activity of PB neurons, leading to amplified responses to sensory stimuli. This mechanism may be related to prolonged catecholaminergic transients.
Article
Neurosciences
Nathan Cramer, Yadong Ji, Maureen A. Kane, Nageswara R. Pilli, Alberto Castro, Luca Posa, Gabrielle Van Patten, Radi Masri, Asaf Keller
Summary: Serotonergic neurons in the rostral ventral medulla play a bidirectional role in pain control. Serotonin release in the spinal cord is pronociceptive, and sustained serotonin signaling may potentiate nociception in chronic pain models.