Oikonomos-II+ is a hybrid, reinforcement-learning system for recommending optimal cloud-instance types for HighPerformance Computing (HPC) and Artificial-Intelligence (AI) applications. Unlike existing approaches that require historical data or repeated job executions, Oikonomos-II+ learns online using user-submitted jobs. It combines a modified Neural-LinUCB algorithm with Gaussian-Process regression to model the relationship between job parameters, instance types, and execution time. This allows it to balance exploration and exploitation efficiently, even in the absence of prior data. We evaluated six configurations of Oikonomos-II+ on a diverse set of HPC and AI workloads, optimizing for cost and speed. Results show that the complete system converges to optimal resource choices, outperforming purely predictive or search-based approaches. By treating deployed applications as a black box and by eliminating the need for preexisting training data or auxiliary runs, Oikonomos-II+ provides a general-purpose, low-overhead solution for dynamic resource selection in heterogeneous cloud environments.

Brain modeling can occur at different levels of abstraction, each aimed at a different purpose. The Virtual Brain (TVB) is an open-source platform for constructing and simulating personalized brain-network models, favoring whole-brain macro-scales while reducing micro-level detail. Among other purposes, TVB is used to build patient-specific, digital, brain twins that can be used in different clinical settings, such as the study and treatment of epilepsy. However, fitting patient-specific TVB models requires a large number of successive and time-consuming simulations. By studying the internal structure of TVB, we observed heterogeneous computation needs in its models which could be leveraged to accelerate simulations. In this work, we designed and implemented HUMA, a heterogeneous, ultra low-latency, dataflow architecture on an AMD Versal Adaptive SoC to accelerate TVB fitting to different patient-brain makeups. Our heterogeneous solution runs about 27× faster compared to a modern-day, server-class, 32-core CPU while consuming a fraction of its power. Additionally, it delivers on average about 14× lower latency, 1.7× better power efficiency and an order-of-magnitude lower energy consumption when compared against the high-performance GPU version of TVB. The achieved latency savings reveal a significant potential in model-fitting for individual patients as well as in closed-loop biohybrid experiments.

Introduction: In 2012, potassium and sodium ion channels in Hodgkin-Huxley-based brain models were shown to exhibit memristive behavior. This positioned memristors as strong candidates for implementing biologically accurate artificial neurons. Memristor-based brain simulations offer advantages in energy efficiency, scalability, and compactness, benefiting fields such as soft robotics, embedded systems, and neuroprosthetics. Methods: Previous approaches used current-controlled Mott memristors, which poorly matched the voltage-controlled nature of ion channels. This study employs volatile, oxide-based memristors that leverage electric-field-driven oxygen-vacancy migration to emulate voltage-dependent channel behavior. We selected candidate WOx and NbOx memristors and modeled their dynamics to verify performance as Hodgkin-Huxley potassium channels. Results: The device exhibits sigmoidal gating and voltage-dependent time constants consistent with the theoretical model. By scaling the passive circuitry around the memristors, we show that they capture the essential mechanisms of potassium ion-channels, although spike height is reduced due to strong non-linear voltage-dependence. Still, by cascading multiple compartments, typical spike propagation is retained. Discussion: This is the first demonstration of a voltage-controlled memristor replicating the Hodgkin-Huxley potassium channel, validating its potential for more efficient brain simulation hardware.

Olivocerebellar learning is highly adaptable, unfolding over minutes to weeks depending on the task. However, the stabilizing mechanisms of the synaptic dynamics necessary for ongoing learning remain unclear. We constructed a model to examine plasticity dynamics under stochastic input and investigate the impact of inferior olive (IO) reverberations on Purkinje cell (PCs) activity and synaptic plasticity. We explored Upbound and Downbound cerebellar micromodules, which are organized loops of IO neurons, cerebellar nuclei neurons and microzones of PCs characterized by their unique molecular profiles and different levels of baseline firing. Our findings show synaptic weight convergence followed by stability of synaptic weights. In line with their relatively low and high intrinsic firing, we observed that Upbound and Downbound PCs have a propensity for potentiation and depression, respectively, with both PC types reaching stability at differential levels of overall strength of their parallel-fiber (PF) inputs. The oscillations and coupling of IO neurons participating in the Upbound and Downbound modules determine at which frequency band PFs can be stabilized optimally. Our results indicate that specific frequency components drive IO resonance and synchronicity, which, in turn, regulate temporal patterning across Upbound and Downbound zones, orchestrating their plasticity dynamics.

Objectives: Current techniques in brain stimulation are still largely based on a phrenologic approach that a single brain target can treat a brain disorder. Nevertheless, meta-analyses of brain implants indicate an overall success rate of 50% improvement in 50% of patients, irrespective of the brain-related disorder. Thus, there is still a large margin for improvement. The goal of this manuscript is to 1) develop a general theoretical framework of brain functioning that is amenable to surgical neuromodulation, and 2) describe the engineering requirements of the next generation of implantable brain stimulators that follow from this theoretic model. Materials and Methods: A neuroscience and engineering literature review was performed to develop a universal theoretical model of brain functioning and dysfunctioning amenable to surgical neuromodulation. Results: Even though a single target can modulate an entire network, research in network science reveals that many brain disorders are the consequence of maladaptive interactions among multiple networks rather than a single network. Consequently, targeting the main connector hubs of those multiple interacting networks involved in a brain disorder is theoretically more beneficial. We, thus, envision next-generation network implants that will rely on distributed, multisite neuromodulation targeting correlated and anticorrelated interacting brain networks, juxtaposing alternative implant configurations, and finally providing solid recommendations for the realization of such implants. In doing so, this study pinpoints the potential shortcomings of other similar efforts in the field, which somehow fall short of the requirements. Conclusion: The concept of network stimulation holds great promise as a universal approach for treating neurologic and psychiatric disorders.

Four-dimensional ultrasound imaging of complex biological systems such as the brain is technically challenging because of the spatiotemporal sampling requirements. We present computational ultrasound imaging (cUSi), an imaging method that uses complex ultrasound fields that can be generated with simple hardware and a physical wave prediction model to alleviate the sampling constraints. cUSi allows for high-resolution four-dimensional imaging of brain hemodynamics in awake and anesthetized mice.

The investigation of neural activity in the murine brain through electrophysiological recordings stands as a fun-damental pursuit within the domain of neuroscience. A specific area of keen interest within this field pertains to the scrutiny of Purkinje cells, nestled within the cerebellum, in order to gain insights into the mechanisms underlying brain injuries and the impairment of motor functions. Notably, Purkinje cells manifest two distinct types of spikes - complex and simple - a pivotal aspect for subsequent classification purposes. However, a critical challenge has persisted in the experimental paradigm: the prevailing setups necessitate the use of wired connections linking the mouse's head stage to data acquisition systems. This constraint substantially curtails the mouse's natural behavior during the course of experimentation, limiting the ability to study essential neural processes and motor function aspects over extended periods. In this paper, we propose a new architectural framework for the detection and classification of neuronal spikes originating from Purkinje cells. This system is engineered to exploit the distinct attributes of these neural entities, effectively winnowing out extraneous data while retaining the pertinent information. The resultant output is a refined dataset, amenable to convenient storage within the mouse's head stage, obviating the need for unwieldy wiring configurations. Our proposed implementation attains a classification accuracy of up to 98% on an in-vivo dataset. Furthermore, its compact form factor en-sures unhindered mobility for the experimental mouse, fostering naturalistic behaviors during the course of scientific inquiry.

In recent years, significant strides in Artificial Intelligence (AI) have led to various practical applications, primarily centered around training and deployment of deep neural networks (DNNs). These applications, however, require considerable computational resources, predominantly reliant on modern Graphics-Processing Units (GPUs). Yet, the quest for larger and faster DNNs has spurred the creation of specialized AI chips and efficient Machine-Learning (ML) software tools like TensorFlow and PyTorch have been developed for striking a balance between usability and performance. Simultaneously, the field of computational neuroscience shares a similar quest for increased computational power to simulate more extensive and detailed brain models, while also keeping usability high. Although GPUs have also entered this field, programming complexity remains high, resulting in cumbersome simulations. Inspired by AI progress, we introduce a workflow for easily accelerating brain simulations using TensorFlow and evaluate the performance of various, cutting-edge AI chips – including the Graphcore Intelligence-Processing Unit (IPU), GroqChip, Nvidia GPU with Tensor Cores, and Google Tensor-Processing Unit (TPU) – when simulating a biologically detailed as well as simpler brain models. Our model simulations explore the architectural tradeoffs of a modern-day CPU and these four AI platforms by varying computational density, memory requirements and floating-point numerical accuracy. Results show that the GroqChip achieves the best performance for small networks, yet is unable to simulate large-scale networks. At the scale of mammalian brains, the GPU, IPU and TPU achieve speedups ranging from 29x to 1,208x times over CPU runtimes. Remarkably, the TPU sets a new record for the largest, real-time simulation of the inferior-olivary nucleus in the brain. Reduced-accuracy floating-point implementations make some simulation results unreliable for brain research, notably for the GroqChip. Consequently, this work underscores the potential of ML libraries for accelerating brain simulations as well as the critical role of AI-chip numerical accuracy for biophysically realistic brain models.

In recent decades, increasing ultrasound frame rates has been the main motivation behind many novel ultrasound imaging applications [1]-[3]. With this work, we propose an efficient ultrafast FPGA beamformer that applies coherent compounding, through a delay-reuse optimization.

The cloud has become a powerful environment for deploying High-Performance Computing (HPC) applications. However, the size and heterogeneity of cloud-hardware offerings poses a challenge in selecting the optimal cloud instance type. Users often lack the knowledge or time necessary to make an optimal choice. In this work, we propose Oikonomos, a data-driven, opportunistic, resource-recommendation system for HPC applications in the cloud. Oikonomos trains a Multi-layer Perceptron (MLP) to predict the performance of a given HPC application, for different input parameters and instance types. It, then, calculates the cost of executing the application on different instance types and proposes the one best-fitting the user's needs. We deployed Oikonomos on a diverse mix of HPC workloads, and found that for all applications, it approached an optimal policy. The optimal instance type was chosen in 90% of the cases for seven out of eight applications, scoring a Mean Absolute Percentage Error (MAPE) consistently below 20%. This demonstrated that Oikonomos can provide a practical, general-purpose, resource-recommendation system for cloud HPC.

Timely detection of cardiac arrhythmia characterized by abnormal heartbeats can help in the early diagnosis and treatment of cardiovascular diseases. Wearable healthcare devices typically use neural networks to provide the most convenient way of continuously monitoring heart activity for arrhythmia detection. However, it is challenging to achieve high accuracy and energy efficiency in these smart wearable healthcare devices. In this work, we provide architecture-level solutions to deploy neural networks for cardiac arrhythmia classification. We have created a hierarchical structure after analyzing various neural network topologies where only required network components are activated to improve energy efficiency while maintaining high accuracy. In our proposed architecture, we introduce a severity-based classification approach to directly help the users of the wearable healthcare device as well as the medical professionals. Additionally, we have employed computation-in-memory based hardware to improve energy efficiency and area consumption by leveraging in-situ data processing and scalability of emerging memory technologies such as resistive random access memory (RRAM). Simulation experiments conducted using the MIT-BIH arrhythmia dataset show that the proposed architecture provides high accuracy while consuming average energy of 0.11 $\mu$J per heartbeat classification and 0.11 mm2 area, thereby achieving 25× improvement in average energy consumption and 12× improvement in area compared to the state-of-the-art.

Mice and rats can rapidly move their whiskers when exploring the environment. Accurate description of these movements is important for behavioral studies in neuroscience. Whisker tracking is, however, a notoriously difficult task due to the fast movements and frequent crossings and juxtapositionings among whiskers. We have recently developed WhiskEras, a computer-vision-based algorithm for whisker tracking in untrimmed, head-restrained mice. Although WhiskEras excels in tracking the movements of individual unmarked whiskers over time based on high-speed videos, the initial version of WhiskEras still had two issues preventing its widespread use: it involved tuning a great number of parameters manually to adjust for different experimental setups, and it was slow, processing less than 1 frame per second. To overcome these problems, we present here WhiskEras 2.0, in which the unwieldy stages of the initial algorithm were improved. The enhanced algorithm is more robust, not requiring intense parameter tuning. Furthermore, it was accelerated by first porting the code from MATLAB to C++ and then using advanced parallelization techniques with CUDA and OpenMP to achieve a speedup of at least 75x when processing a challenging whisker video. The improved WhiskEras 2.0 is made publicly available and is ready for processing high-speed videos, thus propelling behavioral research in neuroscience, in particular on sensorimotor integration.