Quantitative assessment of machine learning reliability and resilience.

Advances in machine learning (ML) have led to applications in safety-critical domains, including security, defense, and healthcare. These ML models are confronted with dynamically changing and actively hostile conditions characteristic of real-world applications, requiring systems incorporating ML to be reliable and resilient. Many studies propose techniques to improve the robustness of ML algorithms. However, fewer consider quantitative techniques to assess changes in the reliability and resilience of these systems over time. To address this gap, this study demonstrates how to collect relevant data during the training and testing of ML suitable for the application of software reliability, with and without covariates, and resilience models and the subsequent interpretation of these analyses. The proposed approach promotes quantitative risk assessment of ML technologies, providing the ability to track and predict degradation and improvement in the ML model performance and assisting ML and system engineers with an objective approach to compare the relative effectiveness of alternative training and testing methods. The approach is illustrated in the context of an image recognition model, which is subjected to two generative adversarial attacks and then iteratively retrained to improve the system's performance. Our results indicate that software reliability models incorporating covariates characterized the misclassification discovery process more accurately than models without covariates. Moreover, the resilience model based on multiple linear regression incorporating interactions between covariates tracks and predicts degradation and recovery of performance best. Thus, software reliability and resilience models offer rigorous quantitative assurance methods for ML-enabled systems and processes.

Tunable and Compartmentalized Multimaterial Bioprinting for Complex Living Tissue Constructs.

Recapitulating inherent heterogeneity and complex microarchitectures within confined print volumes for developing implantable constructs that could maintain their structure in vivo has remained challenging. Here, we present a combinational multimaterial and embedded bioprinting approach to fabricate complex tissue constructs that can be implanted postprinting and retain their three-dimensional (3D) shape in vivo. The microfluidics-based single nozzle printhead with computer-controlled pneumatic pressure valves enables laminar flow-based voxelation of up to seven individual bioinks with rapid switching between various bioinks that can solve alignment issues generated during switching multiple nozzles. To improve the spatial organization of various bioinks, printing fidelity with the z-direction, and printing speed, self-healing and biodegradable colloidal gels as support baths are introduced to build complex geometries. Furthermore, the colloidal gels provide suitable microenvironments like native extracellular matrices (ECMs) for achieving cell growths and fast host cell invasion via interconnected microporous networks in vitro and in vivo. Multicompartment microfibers (i.e., solid, core-shell, or donut shape), composed of two different bioink fractions with various lengths or their intravolume space filled by two, four, and six bioink fractions, are successfully printed in the ECM-like support bath. We also print various acellular complex geometries such as pyramids, spirals, and perfusable branched/linear vessels. Successful fabrication of vascularized liver and skeletal muscle tissue constructs show albumin secretion and bundled muscle mimic fibers, respectively. The interconnected microporous networks of colloidal gels result in maintaining printed complex geometries while enabling rapid cell infiltration, in vivo.

Deep Learning-Enabled Resolution-Enhancement in Mini- and Regular Microscopy for Biomedical Imaging.

Artificial intelligence algorithms that aid mini-microscope imaging are attractive for numerous applications. In this paper, we optimize artificial intelligence techniques to provide clear, and natural biomedical imaging. We demonstrate that a deep learning-enabled super-resolution method can significantly enhance the spatial resolution of mini-microscopy and regular-microscopy. This data-driven approach trains a generative adversarial network to transform low-resolution images into super-resolved ones. Mini-microscopic images and regular-microscopic images acquired with different optical microscopes under various magnifications are collected as our experimental benchmark datasets. The only input to this generative-adversarial-network-based method are images from the datasets down-sampled by the Bicubic interpolation. We use independent test set to evaluate this deep learning approach with other deep learning-based algorithms through qualitative and quantitative comparisons. To clearly present the improvements achieved by this generative-adversarial-network-based method, we zoom into the local features to explore and highlight the qualitative differences. We also employ the peak signal-to-noise ratio and the structural similarity, to quantitatively compare alternative super-resolution methods. The quantitative results illustrate that super-resolution images obtained from our approach with interpolation parameter α=0.25 more closely match those of the original high-resolution images than to those obtained by any of the alternative state-of-the-art method. These results are significant for fields that use microscopy tools, such as biomedical imaging of engineered living systems. We also utilize this generative adversarial network-based algorithm to optimize the resolution of biomedical specimen images and then generate three-dimensional reconstruction, so as to enhance the ability of three-dimensional imaging throughout the entire volumes for spatial-temporal analyses of specimen structures.

A Dynamic Surface Gateway Placement Scheme for Mobile Underwater Networks.

Deployment of surface-level gateways holds potential as an effective method to alleviate high-propagation delays and high-error probability in an underwater wireless sensor network (UWSN). This promise comes from reducing distances to underwater nodes and using radio waves to forward information to a control station. In an UWSN, a dynamic energy efficient surface-level gateway deployment is required to cope with the mobility of underwater nodes while considering the remote and three-dimensional nature of marine space. In general, deployment problems are usually modeled as an optimization problem to satisfy multiple constraints given a set of parameters. One previously published static deployment optimization framework makes assumptions about network workload, routing, medium access control performance, and node mobility. However, in real underwater environments, all these parameters are dynamic. Therefore, the accuracy of performance estimates calculated through static UWSN deployment optimization framework tends to be limited by nature. This paper presents the Prediction-Assisted Dynamic Surface Gateway Placement (PADP) algorithm to maximize the coverage and minimize the average end-to-end delay of a mobile underwater sensor network over a specified period. PADP implements the Interacting Multiple Model (IMM) tracking scheme to predict the positions of sensor nodes. The deployment is determined based on both current and predicted positions of sensor nodes, which enables better coverage and shorter end-to-end delay. PADP uses a branch-and-cut approach to solve the optimization problem efficiently, and employs a disjoint-set data structure to ensure connectivity. Simulation results illustrate that PADP significantly outperforms a static gateway deployment scheme.