Article
Neurosciences
Alberto Patino-Saucedo, Horacio Rostro-Gonzalez, Teresa Serrano-Gotarredona, Bernabe Linares-Barranco
Summary: This work demonstrates that offline-trained Liquid State Machines (LSMs) implemented in the SpiNNaker neuromorphic processor can classify visual events, achieving state-of-the-art performance in the event-based N-MNIST dataset. The readout layer is trained using a variation of back-propagation-through-time (BPTT) for Spiking Neural Networks (SNNs), while the internal weights of the reservoir remain static. Results show that mapping the LSM from a Deep Learning framework to SpiNNaker does not affect the classification task performance. Additionally, weight quantization has a minimal impact on the performance of the LSM.
FRONTIERS IN NEUROSCIENCE
(2022)
Article
Computer Science, Artificial Intelligence
Juan P. Dominguez-Morales, Stefano Buccelli, Daniel Gutierrez-Galan, Ilaria Colombi, Angel Jimenez-Fernandez, Michela Chiappalone
Summary: The accurate identification of burst events is crucial in various fields, yet existing methods in literature are not widely adopted. A novel neuromorphic approach for real-time burst detection proposed in this study shows similar results to current state-of-the-art methods, suggesting potential advantages over conventional techniques.
Article
Engineering, Electrical & Electronic
Ioannis Galanis, Iraklis Anagnostopoulos, Chinh Nguyen, Guillermo Bares
Summary: SNNs have emerged as serious competitors of traditional CNNs, unlocking new potential for more energy-efficient neural networks. A fast exploration framework targeting the SpiNNaker neuromorphic platform has been demonstrated, achieving 98.85% SNN accuracy on the MNIST dataset while reducing exploration time by a factor of 3x.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS II-EXPRESS BRIEFS
(2021)
Article
Neurosciences
Luca Peres, Oliver Rhodes
Summary: This article presents novel multicore processing strategies on the SpiNNaker Neuromorphic hardware, which optimize the efficiency of Spiking Neural Network operations. By parameterizing load balancing between computational units, researchers are able to explore long-term learning and neural pathologies in real- or sub-realtime.
FRONTIERS IN NEUROSCIENCE
(2022)
Article
Neurosciences
Ammar Bitar, Rafael Rosales, Michael Paulitsch
Summary: This paper adapts gradient-based explainability methods originally developed for traditional ANNs to be used in SNNs processing event-based spiking data or real-valued data. The modified methods are evaluated on classification tasks and their accuracy is confirmed through perturbation experiments.
FRONTIERS IN NEUROSCIENCE
(2023)
Article
Mathematical & Computational Biology
Nicolas Skatchkovsky, Hyeryung Jang, Osvaldo Simeone
Summary: This paper takes steps towards designing neuromorphic systems that can adapt to changing learning tasks and provide accurate uncertainty quantification estimates. By deriving online learning rules within a Bayesian continual learning framework, the proposed method updates the distribution parameters of synaptic weights based on prior knowledge and observed data. Experimental results demonstrate the advantages of Bayesian learning over frequentist learning in terms of adaptation and uncertainty quantification.
FRONTIERS IN COMPUTATIONAL NEUROSCIENCE
(2022)
Article
Computer Science, Artificial Intelligence
Pablo Lopez-Osorio, Alberto Patino-Saucedo, Juan P. Dominguez-Morales, Horacio Rostro-Gonzalez, Fernando Perez-Pena
Summary: In recent years, locomotion mechanisms exhibited by vertebrate animals have served as an inspiration for enhancing the performance of robotic systems. This study aims to replicate the adaptability of locomotion seen in vertebrates through a sCPG model. The sCPG generates different rhythmic patterns driven by an external stimulus, allowing the locomotion of a robotic platform to be adapted to the terrain using any sensor as input.
Article
Mathematics
Mingi Jeon, Taewook Kang, Jae-Jin Lee, Woojoo Lee
Summary: This paper focuses on spike-frequency adaptation and proposes a new method with more biological characteristics. The proposed method is shown to significantly reduce the number of spikes while maintaining performance through simulation experiments. Additionally, the paper demonstrates the close relationship between embedding biological meaning in SNNs and the low-power driving characteristics through in-depth analysis.
Article
Multidisciplinary Sciences
Benjamin Cramer, Sebastian Billaudelle, Simeon Kanya, Aron Leibfried, Andreas Grubl, Vitali Karasenko, Christian Pehle, Korbinian Schreiber, Yannik Stradmann, Johannes Weis, Johannes Schemmel, Friedemann Zenke
Summary: This study demonstrates the applicability of surrogate gradient learning on analog neuromorphic hardware using an in the-loop approach. The results show that learning can self-correct for device mismatch and achieve competitive spiking network performance on vision and speech benchmarks. This work sets several benchmarks for low-energy spiking network processing on analog neuromorphic hardware and paves the way for future on-chip learning algorithms.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2022)
Article
Chemistry, Analytical
Jianwei Xue, Lisheng Xie, Faquan Chen, Liangshun Wu, Qingyang Tian, Yifan Zhou, Rendong Ying, Peilin Liu
Summary: Spiking neural networks (SNNs) are ideal for edge computing scenarios due to their intelligent features and energy efficiency. However, current mapping schemes for deploying SNNs onto neuromorphic hardware face limitations such as extended execution times and low throughput. To address these challenges, EdgeMap is introduced as an optimized mapping toolchain specifically designed for efficient deployment of SNNs onto edge devices.
Article
Computer Science, Artificial Intelligence
Luca Zanatta, Alfio Di Mauro, Francesco Barchi, Andrea Bartolini, Luca Benini, Andrea Acquaviva
Summary: This paper presents an energy-efficient implementation of a Reinforcement Learning algorithm using Spiking Neural Networks for obstacle avoidance task on an Unmanned Aerial Vehicle. The SNN algorithm achieves better results than a Convolutional Neural Network, with 6x less energy consumption.
Article
Chemistry, Analytical
Mircea Hulea, George Iulian Uleru, Constantin Florin Caruntu
Summary: The control of anthropomorphic hands should be carried out by artificial units with high biological plausibility, such as adaptive spiking neural networks. These networks can enable robots to learn motions independently through mechanisms like Hebbian learning. This bioinspired concept allows robots to stop their movements based on specific angles without the need for external stimuli.
Article
Neurosciences
Miao Yu, Tingting Xiang, P. Srivatsa, Kyle Timothy Ng Chu, Burin Amornpaisannon, Yaswanth Tavva, Venkata Pavan Kumar Miriyala, Trevor E. Carlson
Summary: Spiking neural networks (SNNs) have the potential to be an efficient alternative to artificial neural networks (ANNs) due to their sparse and low-energy design. However, the accuracy of time-to-first-spike (TTFS) coding in SNNs is not as good as rate-based SNNs. In this study, we propose a novel optimization algorithm and hardware accelerator to improve the accuracy of TTFS-based SNNs and reduce power consumption.
FRONTIERS IN NEUROSCIENCE
(2023)
Article
Chemistry, Analytical
Ling Zhang, Jing Yang, Cong Shi, Yingcheng Lin, Wei He, Xichuan Zhou, Xu Yang, Liyuan Liu, Nanjian Wu
Summary: Neuromorphic hardware systems utilizing spiking neural networks have gained attention for embedded applications due to their brain-inspired, energy-efficient design. This paper introduces a novel VLSI architecture for accelerating deep SCNN inference in real-time low-cost embedded scenarios, achieving high processing speed and recognition accuracies on image datasets. The architecture leverages snapshot processing and fine-grained data pipelines to achieve high throughput, demonstrating the feasibility of SCNN hardware for various embedded applications.
Article
Multidisciplinary Sciences
Te-Yuan Liu, Ata Mahjoubfar, Daniel Prusinski, Luis Stevens
Summary: Neuromorphic computing, which emulates the neural activity of the brain, offers cost and power efficiency advantages. This study applies Intel's Loihi chip to image retrieval and compares its performance with traditional chips. The results demonstrate significant energy efficiency improvements with the neuromorphic solution.
Article
Computer Science, Hardware & Architecture
Luis F. Rojas-Munoz, Santiago Sanchez-Solano, Carlos H. Garcia-Capulin, Horacio Rostro-Gonzalez
Summary: This paper describes a hardware implementation of a genetic algorithm for circle detection in digital images. The results demonstrate the suitability of this proposal for designing embedded systems with restricted size, resources, and energy consumption for applications in the Internet of Things, Industry 4.0, and related paradigms.
COMPUTERS & ELECTRICAL ENGINEERING
(2022)
Article
Computer Science, Artificial Intelligence
Ehsan Rahiminejad, Fatemeh Azad, Adel Parvizi-Fard, Mahmood Amiri, Bernabe Linares-Barranco
Summary: Neurophysiological observations show that the brain can detect and repair impaired synapses, especially through the collaboration of astrocytes retrograde signaling with nearby neurons. A CMOS neuromorphic circuit with self-repairing capabilities has been proposed based on these findings, which can compensate for damaged synapses by modifying signals of healthy synapses. This fault-tolerant circuit is considered a key candidate for future silicon neuronal systems and neuro-inspired circuits.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Neurosciences
Alberto Patino-Saucedo, Horacio Rostro-Gonzalez, Teresa Serrano-Gotarredona, Bernabe Linares-Barranco
Summary: This work demonstrates that offline-trained Liquid State Machines (LSMs) implemented in the SpiNNaker neuromorphic processor can classify visual events, achieving state-of-the-art performance in the event-based N-MNIST dataset. The readout layer is trained using a variation of back-propagation-through-time (BPTT) for Spiking Neural Networks (SNNs), while the internal weights of the reservoir remain static. Results show that mapping the LSM from a Deep Learning framework to SpiNNaker does not affect the classification task performance. Additionally, weight quantization has a minimal impact on the performance of the LSM.
FRONTIERS IN NEUROSCIENCE
(2022)
Article
Computer Science, Artificial Intelligence
Pablo Lopez-Osorio, Alberto Patino-Saucedo, Juan P. Dominguez-Morales, Horacio Rostro-Gonzalez, Fernando Perez-Pena
Summary: In recent years, locomotion mechanisms exhibited by vertebrate animals have served as an inspiration for enhancing the performance of robotic systems. This study aims to replicate the adaptability of locomotion seen in vertebrates through a sCPG model. The sCPG generates different rhythmic patterns driven by an external stimulus, allowing the locomotion of a robotic platform to be adapted to the terrain using any sensor as input.
Article
Engineering, Electrical & Electronic
Jafar Shamsi, Maria Jose Avedillo, Bernabe Linares-Barranco, Teresa Serrano-Gotarredona
Summary: This paper investigates the effect of component mismatches on the performance of differential oscillatory neural networks (DONNs). The results show that mismatches in the components of the differential oscillatory neurons have a greater impact on the performance compared to mismatches in the synaptic circuits. Mismatches in the differential oscillatory neurons result in non-uniformity of the natural frequencies, causing desynchronization and instability.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS I-REGULAR PAPERS
(2023)
Article
Computer Science, Artificial Intelligence
Shuangming Yang, Jiang Wang, Bin Deng, Mostafa Rahimi Azghadi, Bernabe Linares-Barranco
Summary: This study introduces a scalable hardware framework for fault-tolerant context-dependent learning in neuromorphic computing, demonstrating an improvement in network throughput. The proposed system can be utilized for real-time decision-making, brain-machine integration, and research on brain cognition during learning.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Engineering, Electrical & Electronic
Luis A. Camunas-Mesa, Elisa Vianello, Carlo Reita, Teresa Serrano-Gotarredona, Bernabe Linares-Barranco
Summary: This study presents a pseudo-CMOL monolithic chip core that overcomes technical challenges in current CMOS-memristor technologies, using a CMOL-like chip layout technique and a binary weight stochastic STDP learning rule. Experimental results demonstrate successful template matching tasks and feature extraction learning in hardware, achieving perfect recognition.
IEEE JOURNAL ON EMERGING AND SELECTED TOPICS IN CIRCUITS AND SYSTEMS
(2022)
Article
Computer Science, Cybernetics
Amelie Gruel, Dalia Hareb, Antoine Grimaldi, Jean Martinet, Laurent Perrinet, Bernabe Linares-Barranco, Teresa Serrano-Gotarredona
Summary: Foveation refers to the organic action of directing gaze towards a visual region of interest for selective acquisition of relevant information. The recent development of event cameras presents an opportunity to exploit this visual neuroscience mechanism and enhance the efficiency of event data processing. By applying foveation to event data, it is possible to comprehend visual scenes with significantly reduced raw data volume. This study demonstrates the significance of neuromorphic foveation in computer vision tasks such as semantic segmentation and classification, showing a superior trade-off between quantity and quality of information conveyed compared to high or low-resolution event data. The code for this study is publicly available at: https://github.com/amygruel/FoveationStakes_DVS.
BIOLOGICAL CYBERNETICS
(2023)
Article
Computer Science, Information Systems
Morteza Gholami, Gholamreza Karimi, Bernabe Linares-Barranco
Summary: This paper presents an efficient FPGA-based digital implementation of a spiking neuron model and a neuron-astrocyte model. The investigation of astrocytes and noise implications in neural networks are important challenges in neuromorphic studies. The experimental findings show improved hardware synthesis for the proposed models and the role of the approximated astrocyte in neural regulation.
Proceedings Paper
Computer Science, Artificial Intelligence
Adrien Fois, Horacio Rostro-Gonzalez, Bernard Girau
Summary: This study proposes a novel method for unsupervised learning in Spiking Neural Networks (SNN), using spike-timing-dependent plasticity rules (STDP) to adjust synaptic weights and delays for extracting visual representations. Experimental results demonstrate state-of-the-art performances in terms of reconstruction error.
2022 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN)
(2022)
Article
Computer Science, Information Systems
Adan Antonio Alonso-Ramirez, Tat'y Mwata-Velu, Carlos Hugo Garcia-Capulin, Horacio Rostro-Gonzalez, Juan Prado-Olivarez, Marcos Gutierrez-Lopez, Alejandro Israel Barranco-Gutierrez
Summary: This work utilizes two deep learning approaches to classify malaria infected red blood cells from uninfected ones. The proposed deep learning architectures, based on Convolutional-Recurrent neural Networks, achieved an accuracy of 99.89% in detecting malaria-infected red blood cells without preprocessing data, using a malaria's public dataset.
Proceedings Paper
Computer Science, Artificial Intelligence
T. Serrano-Gotarredona, F. Faramarzi, B. Linares-Barranco
Summary: This paper proposes a vision system that uses a foveal mechanism to concentrate attention and dynamically control the center and size of the region of interest. The system is based on an electronically-foveated dynamic vision sensor, and an architecture and implementation for it are presented. Simulation results are provided to demonstrate its operation.
2022 IEEE INTERNATIONAL CONFERENCE ON OMNI-LAYER INTELLIGENT SYSTEMS (IEEE COINS 2022)
(2022)
Proceedings Paper
Automation & Control Systems
Theofilos Spyrou, Sarah A. El-Sayed, Engin Afacan, Luis A. Camunas-Mesa, Bernabe Linares-Barranco, Haralampos-G Stratigopoulos
Summary: This paper assesses the resilience characteristics of a hardware accelerator for Spiking Neural Networks (SNNs) and identifies the parts of the design that need protection against faults and the parts that are inherently fault-tolerant.
PROCEEDINGS OF THE 2022 DESIGN, AUTOMATION & TEST IN EUROPE CONFERENCE & EXHIBITION (DATE 2022)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Amelie Gruel, Jean Martinet, Teresa Serrano-Gotarredona, Bernabe Linares-Barranco
Summary: Event cameras are a new type of sensor that measure pixel-wise changes in brightness and output asynchronous events accordingly. This technology allows for sparse and energy-efficient recording and storage of visual information. Downsizing event data is important for adjusting the complexity of data to available resources, especially in embedded systems.
PROCEEDINGS OF THE 17TH INTERNATIONAL JOINT CONFERENCE ON COMPUTER VISION, IMAGING AND COMPUTER GRAPHICS THEORY AND APPLICATIONS (VISAPP), VOL 4
(2022)
Article
Computer Science, Artificial Intelligence
Hamdan Abdellatef, Lina J. Karam
Summary: This paper proposes performing the learning and inference processes in the compressed domain to reduce computational complexity and improve speed of neural networks. Experimental results show that modified ResNet-50 in the compressed domain is 70% faster than traditional spatial-based ResNet-50 while maintaining similar accuracy. Additionally, a preprocessing step with partial encoding is suggested to improve resilience to distortions caused by low-quality encoded images. Training a network with highly compressed data can achieve good classification accuracy with significantly reduced storage requirements.
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, Artificial Intelligence
Ting Zhou, Hanshu Yan, Jingfeng Zhang, Lei Liu, Bo Han
Summary: We propose a defense strategy that reduces the success rate of data poisoning attacks in downstream tasks by pre-training a robust foundation model.
Article
Computer Science, Artificial Intelligence
Hao Sun, Li Shen, Qihuang Zhong, Liang Ding, Shixiang Chen, Jingwei Sun, Jing Li, Guangzhong Sun, Dacheng Tao
Summary: In this paper, the convergence rate of AdaSAM in the stochastic non-convex setting is analyzed. Theoretical proof shows that AdaSAM has a linear speedup property and decouples the stochastic gradient steps with the adaptive learning rate and perturbed gradient. Experimental results demonstrate that AdaSAM outperforms other optimizers in terms of performance.
Article
Computer Science, Artificial Intelligence
Juntong Yun, Du Jiang, Li Huang, Bo Tao, Shangchun Liao, Ying Liu, Xin Liu, Gongfa Li, Disi Chen, Baojia Chen
Summary: In this study, a dual manipulator grasping detection model based on the Markov decision process is proposed. By parameterizing the grasping detection model of dual manipulators using a cross entropy convolutional neural network and a full convolutional neural network, stable grasping of complex multiple objects is achieved. Robot grasping experiments were conducted to verify the feasibility and superiority of this method.
Article
Computer Science, Artificial Intelligence
Miaohui Zhang, Kaifang Li, Jianxin Ma, Xile Wang
Summary: This paper proposes an unsupervised person re-identification (Re-ID) method that uses two asymmetric networks to generate pseudo-labels for each other by clustering and updates and optimizes the pseudo-labels through alternate training. It also designs similarity compensation and similarity suppression based on the camera ID of pedestrian images to optimize the similarity measure. Extensive experiments show that the proposed method achieves superior performance compared to state-of-the-art unsupervised person re-identification methods.
Article
Computer Science, Artificial Intelligence
Florian Bacho, Dominique Chu
Summary: This paper proposes a new approach called the Forward Direct Feedback Alignment algorithm for supervised learning in deep neural networks. By combining activity-perturbed forward gradients, direct feedback alignment, and momentum, this method achieves better performance and convergence speed compared to other local alternatives to backpropagation.
Article
Computer Science, Artificial Intelligence
Xiaojian Ding, Yi Li, Shilin Chen
Summary: This research paper addresses the limitations of recursive feature elimination (RFE) and its variants in high-dimensional feature selection tasks. The proposed algorithms, which introduce a novel feature ranking criterion and an optimal feature subset evaluation algorithm, outperform current state-of-the-art methods.
Article
Computer Science, Artificial Intelligence
Naoko Koide-Majima, Shinji Nishimoto, Kei Majima
Summary: Visual images observed by humans can be reconstructed from brain activity, and the visualization of arbitrary natural images from mental imagery has been achieved through an improved method. This study provides a unique tool for directly investigating the subjective contents of the brain.
Article
Computer Science, Artificial Intelligence
Huanjie Tao, Qianyue Duan
Summary: In this paper, a hierarchical attention network with progressive feature fusion is proposed for facial expression recognition (FER), addressing the challenges posed by pose variation, occlusions, and illumination variation. The model achieves enhanced performance by aggregating diverse features and progressively enhancing discriminative features.
Article
Computer Science, Artificial Intelligence
Zhenyi Wang, Pengfei Yang, Linwei Hu, Bowen Zhang, Chengmin Lin, Wenkai Lv, Quan Wang
Summary: In the face of the complex landscape of deep learning, we propose a novel subgraph-level performance prediction method called SLAPP, which combines graph and operator features through an innovative graph neural network called EAGAT, providing accurate performance predictions. In addition, we introduce a mixed loss design with dynamic weight adjustment to improve predictive accuracy.
Article
Computer Science, Artificial Intelligence
Yiyang Yin, Shuangling Luo, Jun Zhou, Liang Kang, Calvin Yu-Chian Chen
Summary: Medical image segmentation is crucial for modern healthcare systems, especially in reducing surgical risks and planning treatments. Transanal total mesorectal excision (TaTME) has become an important method for treating colon and rectum cancers. Real-time instance segmentation during TaTME surgeries can assist surgeons in minimizing risks. However, the dynamic variations in TaTME images pose challenges for accurate instance segmentation.
Article
Computer Science, Artificial Intelligence
Teng Cheng, Lei Sun, Junning Zhang, Jinling Wang, Zhanyang Wei
Summary: This study proposes a scheme that combines the start-stop point signal features for wideband multi-signal detection, called Fast Spectrum-Size Self-Training network (FSSNet). By utilizing start-stop points to build the signal model, this method successfully solves the difficulty of existing deep learning methods in detecting discontinuous signals and achieves satisfactory detection speed.
Article
Computer Science, Artificial Intelligence
Wenming Wu, Xiaoke Ma, Quan Wang, Maoguo Gong, Quanxue Gao
Summary: The layer-specific modules in multi-layer networks are critical for understanding the structure and function of the system. However, existing methods fail to accurately characterize and balance the connectivity and specificity of these modules. To address this issue, a joint learning graph clustering algorithm (DRDF) is proposed, which learns the deep representation and discriminative features of the multi-layer network, and balances the connectivity and specificity of the layer-specific modules through joint learning.
Article
Computer Science, Artificial Intelligence
Guanghui Yue, Guibin Zhuo, Weiqing Yan, Tianwei Zhou, Chang Tang, Peng Yang, Tianfu Wang
Summary: This paper proposes a novel boundary uncertainty aware network (BUNet) for precise and robust colorectal polyp segmentation. BUNet utilizes a pyramid vision transformer encoder to learn multi-scale features and incorporates a boundary exploration module (BEM) and a boundary uncertainty aware module (BUM) to handle boundary areas. Experimental results demonstrate that BUNet outperforms other methods in terms of performance and generalization ability.