Search

The promise of large-scale, high-resolution datasets from Electron Microscopy (EM) and X-ray Microtomography (XRM) lies in their ability to reveal neural structures and synaptic connectivity, which is critical for understanding the brain. Effectively managing these complex and rapidly increasing datasets will enable new scientific insights, facilitate querying, and support secondary use across the neuroscience community. However, without effective neurodata standards that permit use of these data across multiple systems and workflows, these valuable and costly datasets risk being underutilized especially as they surpass petascale levels. These standards will promote data sharing through accessible interfaces, allow researchers to build on each other's work, and guide the development of tools and capabilities that are interoperable. Herein we outline a standards framework for creating and managing annotations originating and derived from high-resolution volumetric imaging and connectomic datasets, focusing on ensuring Findable, Accessible, Interoperable, and Reusable (FAIR) practices. The goal is to enhance collaborative efforts, boost the reliability of findings, and enable comparative analysis across growing datasets of different species and modalities. We have formed a global working group with academic and industry partners in the high-resolution volumetric data generation and analysis community, focused on identifying gaps in current EM and XRM data pipelines, and refining outlines and platforms for standardizing EM and XRM methods. This focus considers existing and past community approaches and includes examining neuronal entities, biological components, and associated metadata, while emphasizing adaptability and fostering collaboration., Comment: 11 pages, 4 figures, 1 table

It has been shown that unclocked, recurrent networks of Boolean gates in FPGAs can be used for low-SWaP reservoir computing. In such systems, topology and node functionality of the network are randomly initialized. To create a network that solves a task, weights are applied to output nodes and learning is achieved by adjusting those weights with conventional machine learning methods. However, performance is often limited compared to networks where all parameters are learned. Herein, we explore an alternative learning approach for unclocked, recurrent networks in FPGAs. We use evolutionary computation to evolve the Boolean functions of network nodes. In one type of implementation the output nodes are used directly to perform a task and all learning is via evolution of the network's node functions. In a second type of implementation a back-end classifier is used as in traditional reservoir computing. In that case, both evolution of node functions and adjustment of output node weights contribute to learning. We demonstrate the practicality of node function evolution, obtaining an accuracy improvement of ~30% on an image classification task while processing at a rate of over three million samples per second. We additionally demonstrate evolvability of network memory and dynamic output signals.

The Geiger photodiodes, or single photon avalanche photodiodes, for a radiation tolerant solid-state photomultiplier (SSPM) are being designed using Aluminum Gallium Arsenide (AlGaAs) and are less than 1 micrometer thick. Studies on the changes in dark current compared to silicon from thermal neutron damage were conducted demonstrating that the design concept is feasible, as the device sensitivity to incident radiation is low relative to silicon devices. To achieve the required high fields for Geiger mode operation within a thin device, a triple mesa structure was developed, and simulations show a favorable field profile that constrains the highest electric fields to the middle of the device and keeps the field strength at all corners and surfaces 3-6x lower than the field in the center at breakdown., Comment: 9 pages, 5 figures. Presented at the 2023 Symposium on Radiation Measurements and Applications. To be published in Nuclear Instruments and Methods in Physics Research - Section A

High resolution volumetric neuroimaging datasets from electron microscopy (EM) and x-ray micro and holographic-nano tomography (XRM/XHN) are being generated at an increasing rate and by a growing number of research teams. These datasets are derived from an increasing number of species, in an increasing number of brain regions, and with an increasing number of techniques. Each of these large-scale datasets, often surpassing petascale levels, is typically accompanied by a unique and varied set of metadata. These datasets can be used to derive connectomes, or neuron-synapse level connectivity diagrams, to investigate the fundamental organization of neural circuitry, neuronal development, and neurodegenerative disease. Standardization is essential to facilitate comparative connectomics analysis and enhance data utilization. Although the neuroinformatics community has successfully established and adopted data standards for many modalities, this effort has not yet encompassed EM and XRM/ XHN connectomics data. This lack of standardization isolates these datasets, hindering their integration and comparison with other research performed in the field. Towards this end, our team formed a working group consisting of community stakeholders to develop Image and Experimental Metadata Standards for EM and XRM/XHN data to ensure the scientific impact and further motivate the generation and sharing of these data. This document addresses version 1.1 of these standards, aiming to support metadata services and future software designs for community collaboration. Standards for derived annotations are described in a companion document. Standards definitions are available on a community github page. We hope these standards will enable comparative analysis, improve interoperability between connectomics software tools, and continue to be refined and improved by the neuroinformatics community., Comment: 15 Pages, 3 figures, 2 tables

Scientists are adopting new approaches to scale up their activities and goals. Progress in neurotechnologies, artificial intelligence, automation, and tools for collaboration promises new bursts of discoveries. However, compared to other disciplines and the industry, neuroscience laboratories have been slow to adopt key technologies to support collaboration, reproducibility, and automation. Drawing on progress in other fields, we define a roadmap for implementing automated research workflows for diverse research teams. We propose establishing a five-level capability maturity model for operations in neuroscience research. Achieving higher levels of operational maturity requires new technology-enabled methodologies, which we describe as ``SciOps''. The maturity model provides guidelines for evaluating and upgrading operations in multidisciplinary neuroscience teams., Comment: 10 pages, one figure

We propose a novel modular inference approach combining two different generative models -- generative adversarial networks (GAN) and normalizing flows -- to approximate the posterior distribution of physics-based Bayesian inverse problems framed in high-dimensional ambient spaces. We dub the proposed framework GAN-Flow. The proposed method leverages the intrinsic dimension reduction and superior sample generation capabilities of GANs to define a low-dimensional data-driven prior distribution. Once a trained GAN-prior is available, the inverse problem is solved entirely in the latent space of the GAN using variational Bayesian inference with normalizing flow-based variational distribution, which approximates low-dimensional posterior distribution by transforming realizations from the low-dimensional latent prior (Gaussian) to corresponding realizations of a low-dimensional variational posterior distribution. The trained GAN generator then maps realizations from this approximate posterior distribution in the latent space back to the high-dimensional ambient space. We also propose a two-stage training strategy for GAN-Flow wherein we train the two generative models sequentially. Thereafter, GAN-Flow can estimate the statistics of posterior-predictive quantities of interest at virtually no additional computational cost. The synergy between the two types of generative models allows us to overcome many challenges associated with the application of Bayesian inference to large-scale inverse problems, chief among which are describing an informative prior and sampling from the high-dimensional posterior. We demonstrate the efficacy and flexibility of GAN-Flow on various physics-based inverse problems of varying ambient dimensionality and prior knowledge using different types of GANs and normalizing flows.

Despite the progress in deep learning networks, efficient learning at the edge (enabling adaptable, low-complexity machine learning solutions) remains a critical need for defense and commercial applications. We envision a pipeline to utilize large neuroimaging datasets, including maps of the brain which capture neuron and synapse connectivity, to improve machine learning approaches. We have pursued different approaches within this pipeline structure. First, as a demonstration of data-driven discovery, the team has developed a technique for discovery of repeated subcircuits, or motifs. These were incorporated into a neural architecture search approach to evolve network architectures. Second, we have conducted analysis of the heading direction circuit in the fruit fly, which performs fusion of visual and angular velocity features, to explore augmenting existing computational models with new insight. Our team discovered a novel pattern of connectivity, implemented a new model, and demonstrated sensor fusion on a robotic platform. Third, the team analyzed circuitry for memory formation in the fruit fly connectome, enabling the design of a novel generative replay approach. Finally, the team has begun analysis of connectivity in mammalian cortex to explore potential improvements to transformer networks. These constraints increased network robustness on the most challenging examples in the CIFAR-10-C computer vision robustness benchmark task, while reducing learnable attention parameters by over an order of magnitude. Taken together, these results demonstrate multiple potential approaches to utilize insight from neural systems for developing robust and efficient machine learning techniques., Comment: 11 pages, 4 figures

Optimal sensor placement enhances the efficiency of a variety of applications for monitoring dynamical systems. It has been established that deterministic solutions to the sensor placement problem are insufficient due to the many uncertainties in system input and parameters that affect system response sensor measurements. Accounting for the uncertainties in this typically expensive optimization is challenging due to computational intractability. This study proposes a stochastic environment in the form of a Markov Decision Process and a sensor placement agent that aims to maximize the information gain from placing a particular number of sensors of different types within the system. The agent is trained to maximize its reward based on an information-theoretic reward function. To verify the efficacy, the approach is applied to place a set of heterogeneous sensors in a shear building model. This methodology can be used to accommodate uncertainties in the sensor placement problem in real-world systems.

There are multiple scales of abstraction from which we can describe the same image, depending on whether we are focusing on fine-grained details or a more global attribute of the image. In brain mapping, learning to automatically parse images to build representations of both small-scale features (e.g., the presence of cells or blood vessels) and global properties of an image (e.g., which brain region the image comes from) is a crucial and open challenge. However, most existing datasets and benchmarks for neuroanatomy consider only a single downstream task at a time. To bridge this gap, we introduce a new dataset, annotations, and multiple downstream tasks that provide diverse ways to readout information about brain structure and architecture from the same image. Our multi-task neuroimaging benchmark (MTNeuro) is built on volumetric, micrometer-resolution X-ray microtomography images spanning a large thalamocortical section of mouse brain, encompassing multiple cortical and subcortical regions. We generated a number of different prediction challenges and evaluated several supervised and self-supervised models for brain-region prediction and pixel-level semantic segmentation of microstructures. Our experiments not only highlight the rich heterogeneity of this dataset, but also provide insights into how self-supervised approaches can be used to learn representations that capture multiple attributes of a single image and perform well on a variety of downstream tasks. Datasets, code, and pre-trained baseline models are provided at: https://mtneuro.github.io/ ., Comment: 10 pages, 4 figures, Accepted at NeurIPS 2022

Learning new tasks and skills in succession without losing prior learning (i.e., catastrophic forgetting) is a computational challenge for both artificial and biological neural networks, yet artificial systems struggle to achieve parity with their biological analogues. Mammalian brains employ numerous neural operations in support of continual learning during sleep. These are ripe for artificial adaptation. Here, we investigate how modeling three distinct components of mammalian sleep together affects continual learning in artificial neural networks: (1) a veridical memory replay process observed during non-rapid eye movement (NREM) sleep; (2) a generative memory replay process linked to REM sleep; and (3) a synaptic downscaling process which has been proposed to tune signal-to-noise ratios and support neural upkeep. We find benefits from the inclusion of all three sleep components when evaluating performance on a continual learning CIFAR-100 image classification benchmark. Maximum accuracy improved during training and catastrophic forgetting was reduced during later tasks. While some catastrophic forgetting persisted over the course of network training, higher levels of synaptic downscaling lead to better retention of early tasks and further facilitated the recovery of early task accuracy during subsequent training. One key takeaway is that there is a trade-off at hand when considering the level of synaptic downscaling to use - more aggressive downscaling better protects early tasks, but less downscaling enhances the ability to learn new tasks. Intermediate levels can strike a balance with the highest overall accuracies during training. Overall, our results both provide insight into how to adapt sleep components to enhance artificial continual learning systems and highlight areas for future neuroscientific sleep research to further such systems., Comment: 9 pages, 12 figures, code available upon reasonable request. Corresponding author: William G. Coon (will.coon@jhuapl.edu)

Capturing general deforming scenes from monocular RGB video is crucial for many computer graphics and vision applications. However, current approaches suffer from drawbacks such as struggling with large scene deformations, inaccurate shape completion or requiring 2D point tracks. In contrast, our method, Ub4D, handles large deformations, performs shape completion in occluded regions, and can operate on monocular RGB videos directly by using differentiable volume rendering. This technique includes three new in the context of non-rigid 3D reconstruction components, i.e., 1) A coordinate-based and implicit neural representation for non-rigid scenes, which in conjunction with differentiable volume rendering enables an unbiased reconstruction of dynamic scenes, 2) a proof that extends the unbiased formulation of volume rendering to dynamic scenes, and 3) a novel dynamic scene flow loss, which enables the reconstruction of larger deformations by leveraging the coarse estimates of other methods. Results on our new dataset, which will be made publicly available, demonstrate a clear improvement over the state of the art in terms of surface reconstruction accuracy and robustness to large deformations., Comment: 26 pages, 17 figures, 8 tables

Despite groundbreaking progress in reinforcement learning for robotics, gameplay, and other complex domains, major challenges remain in applying reinforcement learning to the evolving, open-world problems often found in critical application spaces. Reinforcement learning solutions tend to generalize poorly when exposed to new tasks outside of the data distribution they are trained on, prompting an interest in continual learning algorithms. In tandem with research on continual learning algorithms, there is a need for challenge environments, carefully designed experiments, and metrics to assess research progress. We address the latter need by introducing a framework for continual reinforcement-learning development and assessment using Lifelong Learning Explorer (L2Explorer), a new, Unity-based, first-person 3D exploration environment that can be continuously reconfigured to generate a range of tasks and task variants structured into complex and evolving evaluation curricula. In contrast to procedurally generated worlds with randomized components, we have developed a systematic approach to defining curricula in response to controlled changes with accompanying metrics to assess transfer, performance recovery, and data efficiency. Taken together, the L2Explorer environment and evaluation approach provides a framework for developing future evaluation methodologies in open-world settings and rigorously evaluating approaches to lifelong learning., Comment: 10 Pages submitted to AAAI AI for Open Worlds Symposium 2022

We report in-situ synthesis of crystalline indium islands on InAs nanowires grown by molecular beam epitaxy. Structural analysis by transmission electron microscopy showed that In crystals grew in a tetragonal body-centred crystal structure within two families of orientations relative to wurtzite InAs. The crystalline islands had lengths < 500 nm and low-energy surfaces, suggesting that growth was driven mainly by surface energy minimization. Electrical transport through In/InAs devices exhibited Cooper pair charging, evidencing charge parity preservation and a pristine In/InAs interface, with an induced superconducting gap ~ 0.45 meV. Cooper pair charging persisted to temperatures > 1.2 K and magnetic fields ~ 0.7 T, demonstrating that In/InAs hybrids belong to an expanding class of semiconductor/superconductor hybrids operating over a wider parameter space than state-of-the-art Al-based hybrids. Engineering crystal morphology while isolating single islands using shadow epitaxy provides an interesting alternative to previous semiconductor/superconductor hybrid morphologies and device geometries., Comment: Published in Nano Letters: https://pubs.acs.org/doi/10.1021/acs.nanolett.1c02487

State-of-the-art methods for self-supervised learning (SSL) build representations by maximizing the similarity between different transformed "views" of a sample. Without sufficient diversity in the transformations used to create views, however, it can be difficult to overcome nuisance variables in the data and build rich representations. This motivates the use of the dataset itself to find similar, yet distinct, samples to serve as views for one another. In this paper, we introduce Mine Your Own vieW (MYOW), a new approach for self-supervised learning that looks within the dataset to define diverse targets for prediction. The idea behind our approach is to actively mine views, finding samples that are neighbors in the representation space of the network, and then predict, from one sample's latent representation, the representation of a nearby sample. After showing the promise of MYOW on benchmarks used in computer vision, we highlight the power of this idea in a novel application in neuroscience where SSL has yet to be applied. When tested on multi-unit neural recordings, we find that MYOW outperforms other self-supervised approaches in all examples (in some cases by more than 10%), and often surpasses the supervised baseline. With MYOW, we show that it is possible to harness the diversity of the data to build rich views and leverage self-supervision in new domains where augmentations are limited or unknown.