Search

The brain's microvascular cerebral capillary network plays a vital role in maintaining neuronal health, yet capillary dynamics are still not well understood due to limitations in existing imaging techniques. Here, we present Single Capillary Reporters (SCaRe) for transcranial Ultrasound Localization Microscopy (ULM), a novel approach enabling non-invasive, whole-brain mapping of single capillaries and estimates of their transit-time as a neurovascular biomarker. We accomplish this first through computational Monte Carlo and ultrasound simulations of microbubbles flowing through a fully-connected capillary network. We unveil distinct capillary flow behaviors which informs methodological changes to ULM acquisitions to better capture capillaries in vivo. Subsequently, applying SCaRe-ULM in vivo, we achieve unprecedented visualization of single capillary tracks across brain regions, analysis of layer-specific capillary heterogeneous transit times (CHT), and characterization of whole microbubble trajectories from arterioles to venules. Lastly, we evaluate capillary biomarkers using injected lipopolysaccharide to induce systemic neuroinflammation and track the increase in SCaRe-ULM CHT, demonstrating the capability to detect subtle capillary functional changes. SCaRe-ULM represents a significant advance in studying microvascular dynamics, offering novel avenues for investigating capillary patterns in neurological disorders and potential diagnostic applications., Comment: 23 pages, 5 figures

Using drones to track multiple individuals simultaneously in their natural environment is a powerful approach for better understanding group primate behavior. Previous studies have demonstrated that it is possible to automate the classification of primate behavior from video data, but these studies have been carried out in captivity or from ground-based cameras. To understand group behavior and the self-organization of a collective, the whole troop needs to be seen at a scale where behavior can be seen in relation to the natural environment in which ecological decisions are made. This study presents a novel dataset from drone videos for baboon detection, tracking, and behavior recognition. The baboon detection dataset was created by manually annotating all baboons in drone videos with bounding boxes. A tiling method was subsequently applied to create a pyramid of images at various scales from the original 5.3K resolution images, resulting in approximately 30K images used for baboon detection. The tracking dataset is derived from the detection dataset, where all bounding boxes are assigned the same ID throughout the video. This process resulted in half an hour of very dense tracking data. The behavior recognition dataset was generated by converting tracks into mini-scenes, a video subregion centered on each animal; each mini-scene was manually annotated with 12 distinct behavior types, resulting in over 20 hours of data. Benchmark results show mean average precision (mAP) of 92.62\% for the YOLOv8-X detection model, multiple object tracking precision (MOTA) of 63.81\% for the BotSort tracking algorithm, and micro top-1 accuracy of 63.97\% for the X3D behavior recognition model. Using deep learning to classify wildlife behavior from drone footage facilitates non-invasive insight into the collective behavior of an entire group., Comment: Dataset will be published shortly

Various methods have been proposed to enhance image quality in ultrafast ultrasound. Coherent compounding can improve image quality using multiple steered diverging transmits when motion occurring between transmits is corrected for. Harmonic imaging, a standard technique in conventional focused echocardiography, has been adapted for ultrafast imaging, reducing clutter. Coherence-based approaches have also been shown to increase contrast in clinical settings by enhancing signals from coherent echoes and reducing clutter. Herein, we introduce a simple, unified framework that combines motion-correction, harmonic imaging, and angular-coherence, showing for the first time that their benefits can be combined in real-time. Validation was conducted through in vitro testing on a spinning disk model and in vivo on 4 volunteers. In vitro results confirmed the unified framework capability to achieve high contrast in large-motion contexts up to 17 cm/s. In vivo testing highlighted proficiency in generating images of high quality during low and high tissue velocity phases of the cardiac cycle. Specifically, during ventricular filling, the unified framework increased the gCNR from 0.47 to 0.87 when compared against coherent compounding.

Linking cavitation and anatomy was found to be important for predictable outcomes in Focused-Ultrasound Blood-Brain-Barrier-Opening and requires high resolution cavitation mapping. However, cavitation mapping techniques for planning and monitoring of therapeutic procedures either 1) do not leverage the full resolution capabilities of ultrasound imaging or 2) place strong constraints on the length of the therapeutic pulse. This study aimed to develop a high-resolution technique that could resolve vascular anatomy in the cavitation map. Herein, we develop BP-ETACI, derived from bandpass sampling and dual-frequency contrast imaging at 12.5 MHz to produce cavitation maps prior and during blood-brain barrier opening with long therapeutic bursts using a 1.5-MHz focused transducer in the brain of C57BL/6 mice. The BP-ETACI cavitation maps were found to correlate with the vascular anatomy in ultrasound localization microscopy vascular maps and in histological sections. Cavitation maps produced from non-blood-brain-barrier disrupting doses showed the same cavitation-bearing vasculature as maps produced over entire blood-brain-barrier opening procedures, allowing use for 1) monitoring FUS-BBBO, but also for 2) therapy planning and target verification. BP-ETACI is versatile, created high resolution cavitation maps in the mouse brain and is easily translatable to existing FUS-BBBO experiments. As such, it provides a means to further study cavitation phenomena in FUS-BBBO.

Emerging Internet of Things (IoT) and mobile computing applications are expected to support latency-sensitive deep neural network (DNN) workloads. To realize this vision, the Internet is evolving towards an edge-computing architecture, where computing infrastructure is located closer to the end device to help achieve low latency. However, edge computing may have limited resources compared to cloud environments and thus, cannot run large DNN models that often have high accuracy. In this work, we develop REACT, a framework that leverages cloud resources to execute large DNN models with higher accuracy to improve the accuracy of models running on edge devices. To do so, we propose a novel edge-cloud fusion algorithm that fuses edge and cloud predictions, achieving low latency and high accuracy. We extensively evaluate our approach and show that our approach can significantly improve the accuracy compared to baseline approaches. We focus specifically on object detection in videos (applicable in many video analytics scenarios) and show that the fused edge-cloud predictions can outperform the accuracy of edge-only and cloud-only scenarios by as much as 50%. We also show that REACT can achieve good performance across tradeoff points by choosing a wide range of system parameters to satisfy use-case specific constraints, such as limited network bandwidth or GPU cycles., Comment: 12 pages

Three-dimensional (3D) urban models have gained interest because of their applications in many use-cases such as urban planning and virtual reality. However, generating these 3D representations requires LiDAR data, which are not always readily available. Thus, the applicability of automated 3D model generation algorithms is limited to a few locations. In this paper, we propose sat2pc, a deep learning architecture that predicts the point cloud of a building roof from a single 2D satellite image. Our architecture combines Chamfer distance and EMD loss, resulting in better 2D to 3D performance. We extensively evaluate our model and perform ablation studies on a building roof dataset. Our results show that sat2pc was able to outperform existing baselines by at least 18.6%. Further, we show that the predicted point cloud captures more detail and geometric characteristics than other baselines.

Recently over-smoothing phenomenon of Transformer-based models is observed in both vision and language fields. However, no existing work has delved deeper to further investigate the main cause of this phenomenon. In this work, we make the attempt to analyze the over-smoothing problem from the perspective of graph, where such problem was first discovered and explored. Intuitively, the self-attention matrix can be seen as a normalized adjacent matrix of a corresponding graph. Based on the above connection, we provide some theoretical analysis and find that layer normalization plays a key role in the over-smoothing issue of Transformer-based models. Specifically, if the standard deviation of layer normalization is sufficiently large, the output of Transformer stacks will converge to a specific low-rank subspace and result in over-smoothing. To alleviate the over-smoothing problem, we consider hierarchical fusion strategies, which combine the representations from different layers adaptively to make the output more diverse. Extensive experiment results on various data sets illustrate the effect of our fusion method., Comment: Accepted by ICLR 2022 (Spotlight)

Advances in deep vision techniques and ubiquity of smart cameras will drive the next generation of video analytics. However, video analytics applications consume vast amounts of energy as both deep learning techniques and cameras are power-hungry. In this paper, we focus on a parking video analytics platform and propose RL-CamSleep, a deep reinforcement learning-based technique, to actuate the cameras to reduce the energy footprint while retaining the system's utility. Our key insight is that many video-analytics applications do not always need to be operational, and we can design policies to activate video analytics only when necessary. Moreover, our work is complementary to existing work that focuses on improving hardware and software efficiency. We evaluate our approach on a city-scale parking dataset having 76 streets spread across the city. Our analysis demonstrates how streets have various parking patterns, highlighting the importance of an adaptive policy. Our approach can learn such an adaptive policy that can reduce the average energy consumption by 76.38% and achieve an average accuracy of more than 98% in performing video analytics.

While the thermal and electromagnetic properties of plasmonic nanostructures are well understood, nanoscale thermometry still presents an experimental and theoretical challenge. Plasmonic structures can confine electromagnetic energy at the nanoscale, resulting in local, inhomogeneous, controllable heating. But reading out the temperature with nanoscale precision using optical techniques poses a difficult challenge. Here we report on the optical thermometry of individual gold nanorod trimers that exhibit multiple wavelength-dependent plasmon modes resulting in measurably different local temperature distributions. Specifically, we demonstrate how photothermal microscopy encodes different wavelength-dependent temperature profiles in the asymmetry of the photothermal image point spread function. These point spread function asymmetries are interpreted through companion numerical simulations of the photothermal images to reveal how differing thermal gradients within the nanorod trimer can be controlled by exciting its hybridized plasmonic modes. We also find that hybrid plasmon modes that are optically dark can be excited by our focused laser beam illumination geometry at certain beam positions, thereby providing an additional route to modify thermal profiles at the nanoscale beyond wide-field illumination. Taken together these findings demonstrate an all-optical thermometry technique to actively create and measure thermal gradients at the nanoscale below the diffraction limit.

Internet of Things (IoT) devices are becoming increasingly popular and are influencing many application domains such as healthcare and transportation. These devices are used for real-world applications such as sensor monitoring, real-time control. In this work, we look at differentially private (DP) neural network (NN) based network intrusion detection systems (NIDS) to detect intrusion attacks on networks of such IoT devices. Existing NN training solutions in this domain either ignore privacy considerations or assume that the privacy requirements are homogeneous across all users. We show that the performance of existing differentially private stochastic methods degrade for clients with non-identical data distributions when clients' privacy requirements are heterogeneous. We define a cohort-based $(\epsilon,\delta)$-DP framework that models the more practical setting of IoT device cohorts with non-identical clients and heterogeneous privacy requirements. We propose two novel continual-learning based DP training methods that are designed to improve model performance in the aforementioned setting. To the best of our knowledge, ours is the first system that employs a continual learning-based approach to handle heterogeneity in client privacy requirements. We evaluate our approach on real datasets and show that our techniques outperform the baselines. We also show that our methods are robust to hyperparameter changes. Lastly, we show that one of our proposed methods can easily adapt to post-hoc relaxations of client privacy requirements.

We present an approach to data fusion that combines the interpretability of structured probabilistic graphical models with the flexibility of neural networks. The proposed method, lightweight data fusion (LDF), emphasizes posterior analysis over latent variables using two types of information: primary data, which are well-characterized but with limited availability, and auxiliary data, readily available but lacking a well-characterized statistical relationship to the latent quantity of interest. The lack of a forward model for the auxiliary data precludes the use of standard data fusion approaches, while the inability to acquire latent variable observations severely limits direct application of most supervised learning methods. LDF addresses these issues by utilizing neural networks as conjugate mappings of the auxiliary data: nonlinear transformations into sufficient statistics with respect to the latent variables. This facilitates efficient inference by preserving the conjugacy properties of the primary data and leads to compact representations of the latent variable posterior distributions. We demonstrate the LDF methodology on two challenging inference problems: (1) learning electrification rates in Rwanda from satellite imagery, high-level grid infrastructure, and other sources; and (2) inferring county-level homicide rates in the USA by integrating socio-economic data using a mixture model of multiple conjugate mappings.

Recent studies conclude that the global coronavirus (COVID-19) pandemic decreased power sector CO$_2$ emissions globally and in the United States. In this paper, we analyze the statistical significance of CO2 emissions reductions in the U.S. power sector from March through December 2020. We use Gaussian process (GP) regression to assess whether CO2 emissions reductions would have occurred with reasonable probability in the absence of COVID-19 considering uncertainty due to factors unrelated to the pandemic and adjusting for weather, seasonality, and recent emissions trends. We find that monthly CO2 emissions reductions are only statistically significant in April and May 2020 considering hypothesis tests at 5% significance levels. Separately, we consider the potential impact of COVID-19 on coal-fired power plant retirements through 2022. We find that only a small percentage of U.S. coal power plants are at risk of retirement due to a possible COVID-19-related sustained reduction in electricity demand and prices. We observe and anticipate a return to pre-COVID-19 CO2 emissions in the U.S. power sector., Comment: 13 pages, 6 figures, preprint

Buildings consume over 40% of the total energy in modern societies, and improving their energy efficiency can significantly reduce our energy footprint. In this paper, we present \texttt{WattScale}, a data-driven approach to identify the least energy-efficient buildings from a large population of buildings in a city or a region. Unlike previous methods such as least-squares that use point estimates, \texttt{WattScale} uses Bayesian inference to capture the stochasticity in the daily energy usage by estimating the distribution of parameters that affect a building. Further, it compares them with similar homes in a given population. \texttt{WattScale} also incorporates a fault detection algorithm to identify the underlying causes of energy inefficiency. We validate our approach using ground truth data from different geographical locations, which showcases its applicability in various settings. \texttt{WattScale} has two execution modes -- (i) individual, and (ii) region-based, which we highlight using two case studies. For the individual execution mode, we present results from a city containing >10,000 buildings and show that more than half of the buildings are inefficient in one way or another indicating a significant potential from energy improvement measures. Additionally, we provide probable cause of inefficiency and find that 41\%, 23.73\%, and 0.51\% homes have poor building envelope, heating, and cooling system faults, respectively. For the region-based execution mode, we show that \texttt{WattScale} can be extended to millions of homes in the US due to the recent availability of representative energy datasets., Comment: This paper appeared in the Journal ACM Transactions on Data Science

Reducing our reliance on carbon-intensive energy sources is vital for reducing the carbon footprint of the electric grid. Although the grid is seeing increasing deployments of clean, renewable sources of energy, a significant portion of the grid demand is still met using traditional carbon-intensive energy sources. In this paper, we study the problem of using energy storage deployed in the grid to reduce the grid's carbon emissions. While energy storage has previously been used for grid optimizations such as peak shaving and smoothing intermittent sources, our insight is to use distributed storage to enable utilities to reduce their reliance on their less efficient and most carbon-intensive power plants and thereby reduce their overall emission footprint. We formulate the problem of emission-aware scheduling of distributed energy storage as an optimization problem, and use a robust optimization approach that is well-suited for handling the uncertainty in load predictions, especially in the presence of intermittent renewables such as solar and wind. We evaluate our approach using a state of the art neural network load forecasting technique and real load traces from a distribution grid with 1,341 homes. Our results show a reduction of >0.5 million kg in annual carbon emissions -- equivalent to a drop of 23.3% in our electric grid emissions., Comment: 11 pages, 7 figure, This paper will appear in the Proceedings of the ACM International Conference on Future Energy Systems (e-Energy 20) June 2020, Australia

Accurate approximation of the sampling distribution of nonparametric kernel density estimators is crucial for many statistical inference problems. Since these estimators have complex asymptotic distributions, bootstrap methods are often used for this purpose. With i.i.d. observations, a large literature exists concerning optimal bootstrap methods which achieve the fastest possible convergence rate of the bootstrap estimator of the sampling distribution of the kernel density estimator. With dependent data, such an optimality theory is an important open problem. We establish a general theory of optimality of the block bootstrap for kernel density estimation under weak dependence assumptions which are satisfied by many important time series models. We propose a unified framework for a theoretical study of a rich class of bootstrap methods which include as special cases subsampling, Kunsch's moving block bootstrap, Hall's under-smoothing (UNS) as well as approaches incorporating no (NBC) or explicit bias correction (EBC). Moreover, we consider their accuracy under a broad spectrum of choices of the bandwidth $h$, which include as an important special case the MSE-optimal choice, as well as other under-smoothed choices. Under each choice of $h$, we derive the optimal tuning parameters and compare optimal performances between the main subclasses (EBC, NBC, UNS) of the bootstrap methods., Comment: 44 pages

Blockchain technologies are expected to make a significant impact on a variety of industries. However, one issue holding them back is their limited transaction throughput, especially compared to established solutions such as distributed database systems. In this paper, we re-architect a modern permissioned blockchain system, Hyperledger Fabric, to increase transaction throughput from 3,000 to 20,000 transactions per second. We focus on performance bottlenecks beyond the consensus mechanism, and we propose architectural changes that reduce computation and I/O overhead during transaction ordering and validation to greatly improve throughput. Notably, our optimizations are fully plug-and-play and do not require any interface changes to Hyperledger Fabric., Comment: Minor revisions based on reviewer feedback

Cd3As2 is a three-dimensional topological Dirac semimetal with connected Fermi-arc surface states. It has been suggested that topological superconductivity can be achieved in the nontrivial surface states of topological materials by utilizing the superconductor proximity effect. Here we report observations of both {\pi} and 4{\pi} periodic supercurrents in aluminum-Cd3As2-aluminum Josephson junctions. The {\pi} period is manifested by both the magnetic-field dependence of the critical supercurrent and the appearance of half-integer Shapiro steps in the a.c. Josephson effect. Our macroscopic theory suggests that the {\pi} period arises from interference between the induced bulk superconductivity and the induced Fermi arc surface superconductivity. The 4{\pi} period is manifested by the missing first Shapiro steps and is expected for topological superconductivity.

We establish a general theory of optimality for block bootstrap distribution estimation for sample quantiles under a mild strong mixing assumption. In contrast to existing results, we study the block bootstrap for varying numbers of blocks. This corresponds to a hybrid between the subsampling bootstrap and the moving block bootstrap (MBB), in which the number of blocks is somewhere between 1 and the ratio of sample size to block length. Our main theorem determines the optimal choice of the number of blocks and block length to achieve the best possible convergence rate for the block bootstrap distribution estimator for sample quantiles. As part of our analysis, we also prove an important lemma which gives the convergence rate of the block bootstrap distribution estimator, with implications even for the smooth function model. We propose an intuitive procedure for empirical selection of the optimal number and length of blocks. Relevant examples are presented which illustrate the benefits of optimally choosing the number of blocks.

There have been a large number of papers on bismuth ferrite (BiFeO3) over the past few years, trying to exploit its room-temperature magnetoelectric multiferroic properties. Although these are attractive, BiFeO3 is not the ideal multiferroic, due to weak magnetization and the difficulty in limiting leakage currents. Thus there is an ongoing search for alternatives, including such materials as gallium ferrite (GaFeO3). In the present work we report a comprehensive study of the perovskite PbTi1-xPdxO3 with 0 < x < 0.3. Our study includes dielectric, impedance and magnetization measurements, conductivity analysis and study of crystallographic phases present in the samples with special attention paid to minor phases, identified as PdO, PbPdO2, and Pd3Pb. The work is remarkable in two ways: Pd is difficult to substitute into ABO3 perovskite oxides (where it might be useful for catalysis), and Pd is magnetic under only unusual conditions (under strain or internal electric fields). The new material, as a PZT derivative, is expected to have much stronger piezoelectric properties than BiFeO3.

The internal deformation of the brain is far more complex than the rigid motion of the skull. An ultrasound imaging technique that we have developed has a combination of penetration, frame-rate, and motion detection accuracy required to directly observe, for the first time, the formation and evolution of shear shock waves in the brain. Experiments at low impacts on the traumatic brain injury scale demonstrate that they are spontaneously generated and propagate within the porcine brain. Compared to the initially smooth impact, the acceleration at the shock front is amplified up to a factor of 8.5. This highly localized increase in acceleration suggests that shear shock waves are a fundamental mechanism for traumatic injuries in soft tissue.

Applications and systems are constantly faced with decisions that require picking from a set of actions based on contextual information. Reinforcement-based learning algorithms such as contextual bandits can be very effective in these settings, but applying them in practice is fraught with technical debt, and no general system exists that supports them completely. We address this and create the first general system for contextual learning, called the Decision Service. Existing systems often suffer from technical debt that arises from issues like incorrect data collection and weak debuggability, issues we systematically address through our ML methodology and system abstractions. The Decision Service enables all aspects of contextual bandit learning using four system abstractions which connect together in a loop: explore (the decision space), log, learn, and deploy. Notably, our new explore and log abstractions ensure the system produces correct, unbiased data, which our learner uses for online learning and to enable real-time safeguards, all in a fully reproducible manner. The Decision Service has a simple user interface and works with a variety of applications: we present two live production deployments for content recommendation that achieved click-through improvements of 25-30%, another with 18% revenue lift in the landing page, and ongoing applications in tech support and machine failure handling. The service makes real-time decisions and learns continuously and scalably, while significantly lowering technical debt.