Papers by Elishai Ezra Tsur
PLOS Computational Biology, Oct 27, 2022
Biologically plausible computational modeling of visual perception has the potential to link high... more Biologically plausible computational modeling of visual perception has the potential to link high-level visual experiences to their underlying neurons' spiking dynamic. In this work, we propose a neuromorphic (brain-inspired) Spiking Neural Network (SNN)-driven model for the reconstruction of colorful images from retinal inputs. We compared our results to experimentally obtained V1 neuronal activity maps in a macaque monkey using voltage-sensitive dye imaging and used the model to demonstrate and critically explore color constancy, color assimilation, and ambiguous color perception. Our parametric implementation allows critical evaluation of visual phenomena in a single biologically plausible computational framework. It uses a parametrized combination of high and low pass image filtering and SNN-based filling-in Poisson processes to provide adequate color image perception while accounting for differences in individual perception.
Proceedings of the Annual Meeting of the Cognitive Science Society, 2021
Visual perception initiated with a low-level derivation of Spatio-temporal edges and advances to ... more Visual perception initiated with a low-level derivation of Spatio-temporal edges and advances to a higher-level perception of filled surfaces. According to the isomorphic theory, this perceptual filling-in is governed by an activation spread across the retinotopic map, driven from edges to interiors. Here we propose two biologically plausible spiking neural networks, which demonstrate perceptual filling-in by resolving the Poisson equation. Each network exhibits a distinct dynamic and architecture and could be realized and further integrated in the brain.
Frontiers in Neurorobotics
Autonomous driving is one of the hallmarks of artificial intelligence. Neuromorphic (brain-inspir... more Autonomous driving is one of the hallmarks of artificial intelligence. Neuromorphic (brain-inspired) control is posed to significantly contribute to autonomous behavior by leveraging spiking neural networks-based energy-efficient computational frameworks. In this work, we have explored neuromorphic implementations of four prominent controllers for autonomous driving: pure-pursuit, Stanley, PID, and MPC, using a physics-aware simulation framework. We extensively evaluated these models with various intrinsic parameters and compared their performance with conventional CPU-based implementations. While being neural approximations, we show that neuromorphic models can perform competitively with their conventional counterparts. We provide guidelines for building neuromorphic architectures for control and describe the importance of their underlying tuning parameters and neuronal resources. Our results show that most models would converge to their optimal performances with merely 100–1,000 n...
2022 IEEE 21st International Conference on Cognitive Informatics & Cognitive Computing (ICCI*CC)
2022 29th IEEE International Conference on Electronics, Circuits and Systems (ICECS)
2022 IEEE International Conference on Image Processing (ICIP)
Illusionary visual perception has been long used to shed light on biological vision pathways and ... more Illusionary visual perception has been long used to shed light on biological vision pathways and mechanisms. In this work, we propose a biologically plausible spiking neural network with which spike events are used for iterative image reconstruction in which illusionary contrast perception, long known to manifest in human vision, is apparent. This parametric implementation allows us to examine this visual phenomenon in a biologically plausible computational framework, which may also account for differences in individual visual perception.
PLOS Computational Biology
Retinal direction-selectivity originates in starburst amacrine cells (SACs), which display a cent... more Retinal direction-selectivity originates in starburst amacrine cells (SACs), which display a centrifugal preference, responding with greater depolarization to a stimulus expanding from soma to dendrites than to a collapsing stimulus. Various mechanisms were hypothesized to underlie SAC centrifugal preference, but dissociating them is experimentally challenging and the mechanisms remain debatable. To address this issue, we developed the Retinal Stimulation Modeling Environment (RSME), a multifaceted data-driven retinal model that encompasses detailed neuronal morphology and biophysical properties, retina-tailored connectivity scheme and visual input. Using a genetic algorithm, we demonstrated that spatiotemporally diverse excitatory inputs–sustained in the proximal and transient in the distal processes–are sufficient to generate experimentally validated centrifugal preference in a single SAC. Reversing these input kinetics did not produce any centrifugal-preferring SAC. We then explo...
Visual perception initiated with a low-level derivation of Spatio-temporal edges and advances to ... more Visual perception initiated with a low-level derivation of Spatio-temporal edges and advances to a higher-level perception of filled surfaces. According to the isomorphic theory, this perceptual filling-in is governed by an activation spread across the retinotopic map, driven from edges to interiors. Here we propose two biologically plausible spiking neural networks, which demonstrate perceptual filling-in by resolving the Poisson equation. Each network exhibits a distinct dynamic and architecture and could be realized and further integrated in the brain.
Bioinformatics in the Era of Post Genomics and Big Data, 2018
Advancements in integrated neuroscience are often characterized with data-driven approaches for d... more Advancements in integrated neuroscience are often characterized with data-driven approaches for discovery; these progressions are the result of continuous efforts aimed at developing integrated frameworks for the investigation of neuronal dynamics at increasing resolution and in varying scales. Since insights from integrated neuronal models frequently rely on both experimental and computational approaches, simulations and data modeling have inimitable roles. Moreover, data sharing across the neuroscientific community has become an essential component of data-driven approaches to neuroscience as is evident from the number and scale of ongoing national and multinational projects, engaging scientists from diverse branches of knowledge. In this heterogeneous environment, the need to share neuroscientific data as well as to utilize it across different simulation environments drove the momentum for standardizing data models for neuronal morphologies, biophysical properties, and connectivity schemes. Here, I review existing data models in neuroinformatics, ranging from flat to hybrid object-hierarchical approaches, and suggest a framework with which these models can be linked to experimental data, as well as to established records from existing databases. Linking neuronal models and experimental results with data on relevant articles, genes, proteins, disease, etc., might open a new dimension for data-driven neuroscience.
This article is an open access article distributed under the terms and conditions of the Creative... more This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY
2021 IEEE Biomedical Circuits and Systems Conference (BioCAS), 2021
Spike Timing Dependent Plasticity (STDP) is a biologically plausible learning rule routinely used... more Spike Timing Dependent Plasticity (STDP) is a biologically plausible learning rule routinely used for real-time learning in brain-inspired (neuromorphic) systems. In this work, we utilized an analog design of a Neural Engineering Framework (NEF)-tailored spiking neuron, termed OZ, for STDP-driven learning. We propose analog circuit designs of STDP synapse and frequency adaptation and used them to demonstrate longterm potentiation and depression with adapted OZ neurons. Our design provides NEF-compiled energy-efficient STDP with analog circuitry.
2021 IEEE Biomedical Circuits and Systems Conference (BioCAS), 2021
In the past few decades, bioinspired hexapod walking robots have attracted increasing attention, ... more In the past few decades, bioinspired hexapod walking robots have attracted increasing attention, mainly due to their potential to efficiently traverse rough terrains. Recently, neuromorphic (brain-inspired) robotic control has been shown to outperform conventional control paradigms in stochastic environments. In this work, we propose a neuromorphic adaptive body leveling algorithm for a hexapod walking robot during transversal over multi-leveled terrain. We demonstrate adaptive control with distributed accelerator-driven neuro-integrators with only a few thousand spiking neurons. We further propose a framework for the integration of MuJoCo, a modeling environment, and Nengo, a spiking neural networks compiler, for efficient evaluation of neuromorphic control over high degrees of freedom robotic systems in realistic physics-driven scenarios.
2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2021
Event cameras are robust neuromorphic visual sensors, which communicate transients in luminance a... more Event cameras are robust neuromorphic visual sensors, which communicate transients in luminance as events. Current paradigm for image reconstruction from event data relies on direct optimization of artificial Convolutional Neural Networks (CNNs). Here we proposed a two-phase neural network, which comprises a CNN, optimized for Laplacian prediction followed by a Spiking Neural Network (SNN) optimized for Poisson integration. By introducing Laplacian prediction into the pipeline, we provide image reconstruction with a network comprising only 200 parameters. We converted the CNN to SNN, providing a full neuromorphic implementation. We further optimized the network with Mish activation and a novel convoluted CNN design, proposing a hybrid of spiking and artificial neural network with < 100 parameters. Models were evaluated on both N-MNIST and N-Caltech101 datasets.
2021 IEEE Biomedical Circuits and Systems Conference (BioCAS), 2021
One of the first and most remarkable successes in neuromorphic (brain-inspired) engineering was t... more One of the first and most remarkable successes in neuromorphic (brain-inspired) engineering was the development of bio-inspired event cameras, which communicate transients in luminance as events. Here we evaluate the combination of the Channel and Spatial Reliability Tracking (CSRT) algorithm and the LapDepth neural network for the implementation of 3D object tracking with event cameras. We show that following image reconstruction, implemented using the FireNet convolution neural network, visual features are augmented, dramatically increasing tracking performance. We utilized the 3D tracker to neuromorphically represent error-correcting signals. These error-correcting signals can further be used for motion correction in adaptive neurorobotics.
Cognitive Neuroscience, 2021
ABSTRACT Recent findings suggest that electroencephalography (EEG) oscillations in the theta and ... more ABSTRACT Recent findings suggest that electroencephalography (EEG) oscillations in the theta and alpha frequency-bands reflect synchronized interregional neuronal activity and are considered to reflect cognitive-control, and executive working memory mechanisms in humans. Above the age of 50 years, hypothesized pronounced alterations in alpha and theta-band power at resting or across different WM-functioning brain states may well be due to pre-dementia cognitive impairments, or increasing severity of age-related neurological disorders. Executive working memory (EWM) functioning was assessed in older-adult participants (54 to 83 years old) by obtaining their WM-related EEG oscillations and WM performance scores. WM performance and WM brain-state EEG were recorded during online-WM periods as well as during specific online WM events within EWM periods, and during resting offline-WM periods that preceded online-WM periods. Left-prefrontal alpha-power was enhanced during offline-WM periods versus online-WM periods and was significantly related to WM accuracy. Left-prefrontal alpha power and left prefrontal-parietal theta power anterior-posterior difference-gradient during online WM activity were related to reaction times (RT’s). Importantly, during active-storage events, WM-offset offline-periods, and preparatory pre-retrieval events, excessive left-prefrontal alpha activity was related to poor EWM performance. The potential for developing targeted noninvasive cognition-enhancing interventions and developing clinical-monitoring EEG-based biomarkers of pathological cognitive-decline in elderly people is discussed.
Volume 3: Fluid Machinery; Erosion, Slurry, Sedimentation; Experimental, Multiscale, and Numerical Methods for Multiphase Flows; Gas-Liquid, Gas-Solid, and Liquid-Solid Flows; Performance of Multiphase Flow Systems; Micro/Nano-Fluidics, 2018
Integrated microfluidic networks are being rapidly deployed in academia and industry for a vast s... more Integrated microfluidic networks are being rapidly deployed in academia and industry for a vast spectrum of applications, ranging from molecular biology to quantum physics. Current design paradigm for microfluidic layouts is typically based on numerical modeling, which is not suitable for rapid prototyping nor parameter driven design. Here, we utilize the hydraulic-electric circuit analogy to propose a circuit analysis methodology and an open-source framework for a parameter-guided design of integrated microfluidic layouts. We provide a method with which a user can intuitively define the circuit’s constraints and an algorithm which optimizes the hydraulic layout according to physical constraints. Our algorithm supports valves-integrated design and provides a simulation framework that describes fluid flow with different valves configuration.
2018 IEEE 12th International Symposium on Embedded Multicore/Many-core Systems-on-Chip (MCSoC), 2018
Embedded vision processing is currently ingrained into many aspects of modern life, from computer... more Embedded vision processing is currently ingrained into many aspects of modern life, from computer-aided surgeries to navigation of unmanned aerial vehicles. Vision processing can be described using coarse-grained data flow graphs, which were standardized by OpenVX to enable both system and kernel level optimization via separation of concerns. Notably, graph-based specification provides a gateway to a code generation engine, which can produce an optimized, hardware-specific code for deployment. Here we provide an algorithm and JAVA-MVC-based implementation of automated code generation engine for OpenVX-based vision applications, tailored to NVIDIA multiple CUDA Cores SoC Jetson TX. Our algorithm pre-processes the graph, translates it into an ordered layer-oriented data model, and produces C code, which is optimized for the Jetson TX1 and comprised of error checking and iterative execution for real time vision processing.
Annual Review of Biomedical Engineering, 2020
Microfluidic devices developed over the past decade feature greater intricacy, increased performa... more Microfluidic devices developed over the past decade feature greater intricacy, increased performance requirements, new materials, and innovative fabrication methods. Consequentially, new algorithmic and design approaches have been developed to introduce optimization and computer-aided design to microfluidic circuits: from conceptualization to specification, synthesis, realization, and refinement. The field includes the development of new description languages, optimization methods, benchmarks, and integrated design tools. Here, recent advancements are reviewed in the computer-aided design of flow-, droplet-, and paper-based microfluidics. A case study of the design of resistive microfluidic networks is discussed in detail. The review concludes with perspectives on the future of computer-aided microfluidics design, including the introduction of cloud computing, machine learning, new ideation processes, and hybrid optimization.
Advances in Microfluidics - New Applications in Biology, Energy, and Materials Sciences, 2016
Microfluidic applications range from combinatorial chemical synthesis to high-throughput screenin... more Microfluidic applications range from combinatorial chemical synthesis to high-throughput screening, with platforms integrating analog perfusion components, digitally controlled microvalves, and a range of sensors that demand a variety of communication protocols. A comprehensive solution for microfluidic control has to support an arbitrary combination of microfluidic components and to meet the demand for easy-to-operate system as it arises from the growing community of unspecialized microfluidics users. It should also be an easy to modify and extendable platform, which offer an adequate computational resources, preferably without a need for a local computer terminal for increased mobility. Here we will describe several implementation of microfluidics control technologies and propose a microprocessor-based unit that unifies them. Integrated control can streamline the generation process of complex perfusion sequences required for sensor-integrated microfluidic platforms that demand iterative operation procedures such as calibration, sensing, data acquisition, and decision making. It also enables the implementation of intricate optimization protocols, which often require significant computational resources. System integration is an imperative developmental milestone for the field of microfluidics, both in terms of the scalability of increasingly complex platforms that still lack standardization, and the incorporation and adoption of emerging technologies in biomedical research. Here we describe a modular integration and synchronization of a complex multicomponent microfluidic platform.
Uploads
Papers by Elishai Ezra Tsur