Article
Computer Science, Hardware & Architecture
Onkar Susladkar, Gayatri Deshmukh, Subhrajit Nag, Ananya Mantravadi, Dhruv Makwana, Sujitha Ravichandran, R. Sai Chandra Teja, Gajanan H. Chavhan, C. Krishna Mohan, Sparsh Mittal
Summary: In this paper, a novel end-to-end trainable convolutional neural network architecture called ClarifyNet is proposed for single image dehazing. ClarifyNet utilizes low-pass and high-pass filters to extract different types of information and employs a shared-encoder and multi-decoder model structure. By utilizing a weighted loss function, complementary features are extracted and propagated. Experimental results show that ClarifyNet achieves high scores on multiple datasets.
JOURNAL OF SYSTEMS ARCHITECTURE
(2022)
Article
Computer Science, Artificial Intelligence
Sanaullah, Shamini Koravuna, Ulrich Rueckert, Thorsten Jungeblut
Summary: Spiking Neural Networks (SNNs) achieve brain-like efficiency and functionality by mimicking the transmission of electrical signals in the human brain. We introduced the first runtime interactive simulator, RAVSim, which allows end-users to interact, observe outputs, and make changes during simulation. This tool not only increases network design speed but also accelerates user learning capability.
INTERNATIONAL JOURNAL OF NEURAL SYSTEMS
(2023)
Article
Automation & Control Systems
Yu-Hsiu Lee, Shih-Mei Su
Summary: This paper proposes an innovative internal model principle controller based on numerically robust all-pass filters for motion control of galvanometer. The proposed approach allows arbitrary placement of target frequencies and has advantages of lower-order controller and robustness.
CONTROL ENGINEERING PRACTICE
(2023)
Article
Computer Science, Hardware & Architecture
Guoliang Zhu, Xia Hua, Gongjian Yu, Zhilei Chai
Summary: This paper proposes a nonempirical method for predicting the performance of spiking neural network (SNN) simulations, implemented in a hybrid CPU-FPGA cluster. Experimental results show that the method can achieve comparable accuracy without actual simulation runs, with significantly reduced runtime cost.
JOURNAL OF CIRCUITS SYSTEMS AND COMPUTERS
(2022)
Article
Microbiology
Raminta Kazlauskaite, Bachar Cheaib, Chloe Heys, Umer Zeeshan Ijaz, Stephanie Connelly, William Sloan, Julie Russel, Laura Rubio, John Sweetman, Alex Kitts, Philip McGinnity, Philip Lyons, Martin Llewellyn
Summary: This study developed an in vitro gut model (SalmoSim) to simulate the gut compartments of Atlantic salmon and studied the microbial community dynamics and function in response to different dietary formulations. Results showed that SalmoSim microbiomes were stable and representative, with similar responses to new feed as real salmon. The study contributes to the development of in vitro gut systems for fish nutrition and welfare improvement, serving as guidelines for developing similar systems for other fish species.
Article
Polymer Science
Tilen Kosir, Janko Slavic
Summary: Three-dimensional printing by material extrusion allows for the production of fully functional dynamic piezoelectric sensors in a single process, eliminating the need for additional assembly steps. However, these 3D-printed sensors have higher electrical resistance due to material limitations, resulting in a limited usable frequency range. This research presents an analytical model for determining the usable frequency range of 3D-printed piezoelectric sensors with resistive electrodes, and experimental results confirm the accuracy of the model. This research opens up possibilities for designing the electrical and dynamic characteristics of future intelligent dynamic systems printed in a single process.
Article
Construction & Building Technology
Zhexiong Shang, Zhigang Shen
Summary: This article introduces a new online inspection method that utilizes depth camera array to create a 3D pipeline model. By applying a new camera calibration method and robust camera motion estimation approach, high-precision 3D pipeline surface reconstruction is achieved. The experimental results demonstrate the high accuracy and potential application value of this method.
AUTOMATION IN CONSTRUCTION
(2022)
Article
Biochemistry & Molecular Biology
Andrei L. Lomize, Kevin A. Schnitzer, Spencer C. Todd, Stanislav Cherepanov, Carlos Outeiral, Charlotte M. Deane, Irina D. Pogozheva
Summary: The new version of the Membranome database, Membranome 3.0, has been significantly updated by incorporating models generated by AlphaFold 2. It provides comprehensive structural information on membrane proteins from different organisms and includes new tools for visualization and analysis.
Article
Energy & Fuels
Jianbo Zhang, Longqiao Chen, Shaowei Pan, Hui Liu, Yongle Ma, Jihao Pei, Baojiang Sun, Zhiyuan Wang
Summary: This study analyzed the formation of hydrate plugging in single-pass gas-dominant flows in real time and found that a sudden increase in pressure drop, a sudden drop in temperature, and gas flow rate were important signals indicating hydrate plugging formation. As the subcooling of hydrate formation increased, the hydrate deposition ratio increased gradually, and the number of sloughing events and the time required for hydrate plugging significantly decreased. Based on these findings, a formation mechanism and early warning method for hydrate plugging in single-pass gas-dominant flows were proposed. This study provides valuable insights for the monitoring and prevention of hydrate plugging during natural gas field development.
Article
Mathematics, Applied
Lois Naudin
Summary: The goal of this paper is to better understand the emergence of non-spiking neuron behavior through the analysis of mathematical models. The study suggests that the up-and down-states in non-spiking neurons may be driven by bistable presynaptic neurons, rather than the intrinsic properties of the neurons themselves.
Article
Computer Science, Artificial Intelligence
Chih-Hsu Huang, Chou-Ching K. Lin
Summary: The density-based neural mass model (dNMM) is a novel approach to model the dynamics of adaptive exponential integrate-and-fire neurons, capturing essential neuronal features such as voltage-dependent conductance-based synaptic interactions and adaptation of firing rate responses. It accurately estimates firing rate responses of neuronal populations to different inputs and describes the impact of spike-frequency adaptation on the generation of asynchronous irregular activity in excitatory-inhibitory cortical networks. The dNMM is a promising candidate for building large-scale network models involving multiple brain areas due to its biological realism and computational efficiency.
Article
Energy & Fuels
Qingqing He, Lei Liu, Mingyang Qiu, Quanming Luo
Summary: This paper demonstrates the design method of a passive low-pass filter for a single-stage high-frequency AC/AC converter, ensuring the requirements of power factor and harmonics are met through parameter analysis and design. After validation through PSIM simulation and a 130-W prototype, the effectiveness and reliability of the design method are confirmed.
Article
Engineering, Electrical & Electronic
Milad Ghanbarpour, Ali Naderi, Saeed Haghiri, Arash Ahmadi
Summary: This study focuses on low-cost hardware implementation for retinal cone cells. The proposed model uses multi-linear functions to approximate nonlinear terms and eliminate multiplication expressions, achieving high precision behavior tracking and good match in dynamic behaviors with the main model. Hardware implementation results show that the proposed model is fully valid, with lower hardware volume, 4 times higher frequency, and 22% less power consumption than the original model.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS I-REGULAR PAPERS
(2021)
Article
Engineering, Marine
Seyed Masoud Mahmoudof, Seyed Mostafa Siadatmousavi, Mohammadali Lotfi Takami
Summary: This study primarily investigates the capability of the one-dimensional mode of the SWASH model to reproduce bound super-harmonics resulted from self-coupling of the spectral peak. The model results were compared with wave measurements, and it was found that the model can properly reproduce nonlinear waves experiencing shoaling and breaking. The model also simulated secondary and tertiary local peaks as results of self-coupling.
Article
Engineering, Electrical & Electronic
Samir Gautam, Weidong Xiao, Dylan Dah-Chuan Lu, Hafiz Ahmed, Josep M. Guerrero
Summary: This article presents a new solution for achieving PLL function in single-phase grid interconnection and eliminating additional frequency feedback loops in traditional PLL architecture. Four new topologies are developed and compared for their dynamic response, steady-state accuracy, implementation, and disturbance rejection capability.
IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS
(2022)
Review
Neurosciences
Eduardo Perez-Valero, Miguel A. Lopez-Gordo, Christian Morillas, Francisco Pelayo, Miguel A. Vaquero-Blasco
Summary: This paper reviews state-of-the-art approaches using signal processing and machine learning to automate the detection of Alzheimer's disease and its prodromal stages, focusing on EEG as a suitable alternative for early detection and summarizing the current research on automatic detection.
JOURNAL OF ALZHEIMERS DISEASE
(2021)
Article
Computer Science, Artificial Intelligence
Francisco Barranco, Cornelia Fermueller, Yiannis Aloimonos, Eduardo Ros
Summary: This paper investigates the impact of avoiding optical flow estimation on structure recovery, and proposes a new method based on image gradients to solve 3D motion problems by reformulating the positive-depth constraint. Experimental results show that the method achieves higher accuracy and outperforms existing techniques based on normal flow for 3D motion estimation.
PATTERN RECOGNITION
(2021)
Article
Chemistry, Analytical
Miguel A. Vaquero-Blasco, Eduardo Perez-Valero, Christian Morillas, Miguel A. Lopez-Gordo
Summary: The latest studies have shown that 360-degree VR experiences can significantly reduce stress, reduce costs, and make stress relief assistance more accessible to the general public, such as in workplaces or homes.
Article
Biochemical Research Methods
Pablo Martinez-Canada, Torbjorn V. Ness, Gaute T. Einevoll, Tommaso Fellin, Stefano Panzeri
Summary: This study derived a new mathematical expression called EEG proxy, which can accurately estimate EEG signals based on simulations of point-neuron network models. The new proxies outperformed previous approaches and provided a better approximation of EEG spectra and evoked potentials. This work offers important mathematical tools for interpreting experimentally measured EEGs within neural models of brain function.
PLOS COMPUTATIONAL BIOLOGY
(2021)
Article
Automation & Control Systems
Ignacio Abadia, Francisco Naveros, Jesus A. Garrido, Eduardo Ros, Niceto R. Luque
Summary: The work presents a novel biological approach for real-time control of a robotic arm, integrating a spiking cerebellar network to provide torque-driven control for accurate and coordinated movements. The spiking cerebellar controller uses sensorial signals, goal behavior, and an instructive signal to compute torque commands, supporting spike-timing-dependent plasticity (STDP) for adaptive control. This compliant approach outperforms factory-installed position control in tasks involving cerebellar motor behavior, including smooth movements, fast ballistic movements, and unstructured scenario movements.
IEEE TRANSACTIONS ON CYBERNETICS
(2021)
Article
Mathematical & Computational Biology
Eduardo Perez-Valero, Miguel A. Vaquero-Blasco, Miguel A. Lopez-Gordo, Christian Morillas
Summary: The study introduces a quantitative stress assessment method based on EEG and regression algorithms, which predicts participants' stress levels with remarkable performance. These results could have a positive impact in fields like neuromarketing and professional training for individuals facing stressful situations.
FRONTIERS IN COMPUTATIONAL NEUROSCIENCE
(2021)
Article
Robotics
Ignacio Abadia, Francisco Naveros, Eduardo Ros, Richard R. Carrillo, Niceto R. Luque
Summary: This study aims to bridge the inherent features of cerebellar motor control and current robotic challenges by implementing a cerebellar-like spiking neural network (SNN) controller that is adaptive, compliant, and robust to variable sensorimotor delays. By replicating the cerebellar mechanisms and allowing motor learning and adaptation, this controller addresses the incompatibility between HRI robots' characteristics and traditional control solutions, as well as the challenge of compliant control in the presence of variable sensorimotor delays.
Article
Computer Science, Artificial Intelligence
Niceto R. Luque, Francisco Naveros, Denis Sheynikhovich, Eduardo Ros, Angelo Arleo
Summary: The study explores the relationship between cerebellum-dependent VOR adaptation and structural and functional changes during aging. It finds that long-term plasticity and intrinsic plasticity play important roles in maintaining stable VOR function. The research also highlights the value of computational epidemiology in understanding discrepancies in human cross-sectional studies, as well as the predictive significance of residual fiber quantity in encoding the peak and trough of VOR trajectories.
Article
Computer Science, Interdisciplinary Applications
Eduardo Perez-Valero, Miguel Angel Lopez-Gordo, Christian Morillas Gutierrez, Ismael Carrera-Munoz, Rosa M. Vilchez-Carrillo
Summary: Early detection is crucial for controlling Alzheimer's disease and delaying cognitive decline. Researchers have evaluated AD detection methods using machine learning and EEG. This study presents a preliminary evaluation of a self-driven AD multi-class discrimination approach based on commercial EEG and machine learning, showing the potential for AD detection through this self-driven approach.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2022)
Article
Mathematical & Computational Biology
Eduardo Perez-Valero, Christian Morillas, Miguel A. Lopez-Gordo, Ismael Carrera-Munoz, Samuel Lopez-Alcalde, Rosa M. Vilchez-Carrillo
Summary: Early detection of Alzheimer's disease is crucial and current techniques are costly or invasive. Researchers have investigated AD detection using electroencephalography and machine learning algorithms. They performed a preliminary evaluation using a commercial EEG system and automated classification pipeline. The results suggest that AD can be automatically detected using this approach, which could potentially reduce costs and shorten detection times.
FRONTIERS IN NEUROINFORMATICS
(2022)
Article
Computer Science, Artificial Intelligence
Eduardo Perez-Valero, Christian Morillas, Miguel A. Lopez-Gordo, Jesus Minguillon
Summary: Alzheimer's disease (AD) is the most common form of dementia that lacks a cure, but medical treatment can slow its progression. Early-stage diagnosis is crucial for improving the living standards of patients, but existing diagnostic techniques are limited. In this study, the feasibility of using a reduced four-channel EEG montage for early-stage AD detection was evaluated. Results showed similar accuracies compared to a 16-channel montage, suggesting that a four-channel wearable EEG system could effectively support early-stage AD detection.
INTERNATIONAL JOURNAL OF NEURAL SYSTEMS
(2023)
Article
Engineering, Biomedical
Eduardo Perez-Valero, Christian A. Morillas Gutierrez, Miguel Angel Lopez-Gordo, Samuel Lopez Alcalde
Summary: In this study, the potential of a visual dynamics evaluation called RSVP for detecting cognitive impairment in Alzheimer's disease (AD) was analyzed and compared with two commonly used cognitive tests in Spain. The preliminary results showed that the RSVP performance was significantly better in the control group compared to the patients, and the test exhibited high classification accuracy. This test can contribute to speeding up cognitive impairment screening and reducing associated costs.
JOURNAL OF NEUROENGINEERING AND REHABILITATION
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Pablo Martinez-Canada, Stefano Panzeri
Summary: This study utilized simulations to investigate inferring neural circuit parameters from microscale neural electrical activity, explored the relationship between different spectral features and excitatory-inhibitory balance in neural circuits, and outlined plans to fit the model to empirical measurements of neural activity in future work.
BRAIN INFORMATICS, BI 2021
(2021)
Article
Computer Science, Information Systems
Francisco Girela-Lopez, Eduardo Ros, Javier Diaz
Summary: Network visibility and monitoring are crucial in modern networks, especially in sectors like telecom, power grids, and finance. This paper introduces a novel network time monitoring mechanism that utilizes picosecond precision packet timestamping and White Rabbit synchronization protocol to provide unprecedented packet capturing correlation accuracy in distributed network scenarios.
Article
Computer Science, Interdisciplinary Applications
M. A. Lopez-Gordo, Nico Kohlmorgen, C. Morillas, Francisco Pelayo
Summary: Researchers have focused on high-level analysis of win/lose chances and player performance in the video gaming industry, but there has not been prediction at the single-action level for games like first-person shooters.