Search

Sustaining high fidelity and high throughput of perception tasks over vision sensor streams on edge devices remains a formidable challenge, especially given the continuing increase in image sizes (e.g., generated by 4K cameras) and complexity of DNN models. One promising approach involves criticality-aware processing, where the computation is directed selectively to critical portions of individual image frames. We introduce MOSAIC, a novel system for such criticality-aware concurrent processing of multiple vision sensing streams that provides a multiplicative increase in the achievable throughput with negligible loss in perception fidelity. MOSAIC determines critical regions from images received from multiple vision sensors and spatially bin-packs these regions using a novel multi-scale Mosaic Across Scales (MoS) tiling strategy into a single canvas frame, sized such that the edge device can retain sufficiently high processing throughput. Experimental studies using benchmark datasets for two tasks, Automatic License Plate Recognition and Drone-based Pedestrian Detection, show that MOSAIC, executing on a Jetson TX2 edge device, can provide dramatic gains in the throughput vs. fidelity tradeoff. For instance, for drone-based pedestrian detection, for a batch size of 4, MOSAIC can pack input frames from 6 cameras to achieve (a) 4.75x higher throughput (23 FPS per camera, cumulatively 138FPS) with less than 1% accuracy loss, compared to a First Come First Serve (FCFS) processing paradigm., Comment: To appear in ACM Multimedia Systems 2023

The proliferation of political memes in modern information campaigns calls for efficient solutions for image classification by ideological affiliation. While significant advances have recently been made on text classification in modern natural language processing literature, understanding the political insinuation in imagery is less developed due to the hard nature of the problem. Unlike text, where meaning arises from juxtaposition of tokens (words) within some common linguistic structures, image semantics emerge from a much less constrained process of fusion of visual concepts. Thus, training a model to infer visual insinuation is possibly a more challenging problem. In this paper, we explore an alternative unsupervised approach that, instead, infers ideological affiliation from image propagation patterns on social media. The approach is shown to improve the F1-score by over 0.15 (nearly 25%) over previous unsupervised baselines, and then by another 0.05 (around 7%) in the presence of a small amount of supervision., Comment: n Proc. PhoMemes (in conjunction with ICWSM), Limassol, Cyprus, June 2023

Purely data-driven deep neural networks (DNNs) applied to physical engineering systems can infer relations that violate physics laws, thus leading to unexpected consequences. To address this challenge, we propose a physics-model-based DNN framework, called Phy-Taylor, that accelerates learning compliant representations with physical knowledge. The Phy-Taylor framework makes two key contributions; it introduces a new architectural Physics-compatible neural network (PhN), and features a novel compliance mechanism, we call {\em Physics-guided Neural Network Editing\}. The PhN aims to directly capture nonlinearities inspired by physical quantities, such as kinetic energy, potential energy, electrical power, and aerodynamic drag force. To do so, the PhN augments neural network layers with two key components: (i) monomials of Taylor series expansion of nonlinear functions capturing physical knowledge, and (ii) a suppressor for mitigating the influence of noise. The neural-network editing mechanism further modifies network links and activation functions consistently with physical knowledge. As an extension, we also propose a self-correcting Phy-Taylor framework that introduces two additional capabilities: (i) physics-model-based safety relationship learning, and (ii) automatic output correction when violations of safety occur. Through experiments, we show that (by expressing hard-to-learn nonlinearities directly and by constraining dependencies) Phy-Taylor features considerably fewer parameters, and a remarkably accelerated training process, while offering enhanced model robustness and accuracy., Comment: Working Paper

Object detection in state-of-the-art Autonomous Vehicles (AV) framework relies heavily on deep neural networks. Typically, these networks perform object detection uniformly on the entire camera LiDAR frames. However, this uniformity jeopardizes the safety of the AV by giving the same priority to all objects in the scenes regardless of their risk of collision to the AV. In this paper, we present a new end-to-end pipeline for AV that introduces the concept of LiDAR cluster first and camera inference later to detect and classify objects. The benefits of our proposed framework are twofold. First, our pipeline prioritizes detecting objects that pose a higher risk of collision to the AV, giving more time for the AV to react to unsafe conditions. Second, it also provides, on average, faster inference speeds compared to popular deep neural network pipelines. We design our framework using the real-world datasets, the Waymo Open Dataset, solving challenges arising from the limitations of LiDAR sensors and object detection algorithms. We show that our novel object detection pipeline prioritizes the detection of higher risk objects while simultaneously achieving comparable accuracy and a 25% higher average speed compared to camera inference only., Comment: 6 pages

This paper reviews the novel concept of controllable variational autoencoder (ControlVAE), discusses its parameter tuning to meet application needs, derives its key analytic properties, and offers useful extensions and applications. ControlVAE is a new variational autoencoder (VAE) framework that combines the automatic control theory with the basic VAE to stabilize the KL-divergence of VAE models to a specified value. It leverages a non-linear PI controller, a variant of the proportional-integral-derivative (PID) control, to dynamically tune the weight of the KL-divergence term in the evidence lower bound (ELBO) using the output KL-divergence as feedback. This allows us to precisely control the KL-divergence to a desired value (set point), which is effective in avoiding posterior collapse and learning disentangled representations. In order to improve the ELBO over the regular VAE, we provide simplified theoretical analysis to inform setting the set point of KL-divergence for ControlVAE. We observe that compared to other methods that seek to balance the two terms in VAE's objective, ControlVAE leads to better learning dynamics. In particular, it can achieve a good trade-off between reconstruction quality and KL-divergence. We evaluate the proposed method on three tasks: image generation, language modeling and disentangled representation learning. The results show that ControlVAE can achieve much better reconstruction quality than the other methods for comparable disentanglement. On the language modeling task, ControlVAE can avoid posterior collapse (KL vanishing) and improve the diversity of generated text. Moreover, our method can change the optimization trajectory, improving the ELBO and the reconstruction quality for image generation., Comment: arXiv admin note: substantial text overlap with arXiv:2004.05988

As the ongoing rapid urbanization takes place with an ever-increasing speed, fully modeling urban dynamics becomes more and more challenging, but also a necessity for socioeconomic development. It is challenging because human activities and constructions are ubiquitous; urban landscape and life content change anywhere and anytime. It's crucial due to the fact that only up-to-date urban dynamics can enable governors to optimize their city planning strategy and help individuals organize their daily lives in a more efficient way. Previous geographic topic model based methods attempt to solve this problem but suffer from high computational cost and memory consumption, limiting their scalability to city level applications. Also, strong prior assumptions make such models fail to capture certain patterns by nature. To bridge the gap, we propose Macross, a metapath guided embedding approach to jointly model location, time and text information. Given a dataset of geo-tagged social media posts, we extract and aggregate location and time and construct a heterogeneous information network using the aggregated space and time. Metapath2vec based approach is used to construct vector representations for times, locations and frequent words such that co-occurrence pairs of nodes are closer in latent space. The vector representations will be used to infer related time, locations or keywords for a user query. Experiments done on enormous datasets show our model can generate comparable if not better quality query results compared to state of the art models and outperform some cutting-edge models for activity recovery and classification.

We explore the resiliency and robustness of systems while viewing them as complex, multi-genre networks. The term "complex, multi-genre networks" refers to networks that combine several distinct genres - networks of physical resources, communication networks, information networks, and social and cognitive networks. We show that this perspective is fruitful and adds to our understanding of fundamental challenges and tradeoffs in robustness and resiliency, as well as potential solutions to the challenges. Study of systems as multi-genre networks is relatively uncommon; instead, it is customary in research and engineering literature to focus on a view of a network comprised of homogeneous elements, (e.g., a network of communication devices, or a network of social beings). Yet, most if not all real-world networks are multi-genre - it is hard to find any real system of a significant complexity that does not include a combination of interconnected physical elements, communication devices and channels, data collections, and human users forming an integrated, inter-dependent whole. Most approaches to improving resiliency and robustness involve compromises, and the key challenge is to find a favorable compromise. Such compromises involve reducing or managing the complexity of the network: coupling, rigidity and dependency. We discuss several of these compromises, e.g., performance vs resiliency; resiliency to one type of disruption vs resiliency to another disruption type; and complexity vs resiliency., Comment: A version of this paper appeared as a book chapter in Adaptive, Dynamic, and Resilient Systems published by Springer

© 2017 ACM. The increased penetration of uncertain and variable renewable energy presents various resource and operational electric grid challenges 1 . Micro-level (household and small commercial) demand-side grid flexibility could be a cost-effective strategy to integrate high penetrations of wind and solar energy, but literature and field deployments exploring the necessary information and communication technologies (ICTs) are scant. This paper presents an exploratory framework for enabling information driven grid flexibility through the Internet of Things (IoT), and a proof-of-concept wireless sensor gateway (FlexBox) to collect the necessary parameters for adequately monitoring and actuating the micro-level demand-side. In the summer of 2015, thirty sensor gateways were deployed in the city of Managua (Nicaragua) to develop a baseline for a near future small-scale demand response pilot implementation. FlexBox field data has begun shedding light on relationships between ambient temperature and load energy consumption, load and building envelope energy efficiency challenges, latency communication network challenges, and opportunities to engage existing demand-side user behavioral patterns. Information driven grid flexibility strategies present great opportunity to develop new technologies, system architectures, and implementation approaches that can easily scale across regions, incomes, and levels of development.