Search for: All records

Mobile and embedded devices are becoming ubiquitous. Applications such as rescue with autonomous robots and event analysis on traffic cameras rely on devices with limited power supply and computational sources. Thus, the demand for efficient computer vision algorithms increases. Since 2015, we have organized the IEEE Low-Power Computer Vision Challenge to advance the state of the art in low-power computer vision. We describe the competition organizing details including the challenge design, the reference solution, the dataset, the referee system, and the evolution of the solutions from two winning teams. We examine the winning teams’ development patterns and design decisions, focusing on their techniques to balance power consumption and accuracy. We conclude that a successful competition needs a well-designed reference solution and automated referee system, and a solution with modularized components is more likely to win. We hope this paper provides guidelines for future organizers and contestants of computer vision competitions. 

Deep Neural Networks (DNNs) are being adopted as components in software systems. Creating and specializing DNNs from scratch has grown increasingly difficult as stateof- the-art architectures grow more complex. Following the path of traditional software engineering, machine learning engineers have begun to reuse large-scale pre-trained models (PTMs) and fine-tune these models for downstream tasks. Prior works have studied reuse practices for traditional software packages to guide software engineers towards better package maintenance and dependency management. We lack a similar foundation of knowledge to guide behaviors in pre-trained model ecosystems. In this work, we present the first empirical investigation of PTM reuse. We interviewed 12 practitioners from the most popular PTM ecosystem, Hugging Face, to learn the practices and challenges of PTM reuse. From this data, we model the decision-making process for PTM reuse. Based on the identified practices, we describe useful attributes for model reuse, including provenance, reproducibility, and portability. Three challenges for PTM reuse are missing attributes, discrepancies between claimed and actual performance, and model risks. We substantiate these identified challenges with systematic measurements in the Hugging Face ecosystem. Our work informs future directions on optimizing deep learning ecosystems by automated measuring useful attributes and potential attacks, and envision future research on infrastructure and standardization for model registries. 

Abstract A discrete degree of freedom can be engineered to match the Hamiltonian of particles moving in a real-space lattice potential. Such synthetic dimensions are powerful tools for quantum simulation because of the control they offer and the ability to create configurations difficult to access in real space. Here, in an ultracold 84 Sr atom, we demonstrate a synthetic-dimension based on Rydberg levels coupled with millimeter waves. Tunneling amplitudes between synthetic lattice sites and on-site potentials are set by the millimeter-wave amplitudes and detunings respectively. Alternating weak and strong tunneling in a one-dimensional configuration realizes the single-particle Su-Schrieffer-Heeger (SSH) Hamiltonian, a paradigmatic model of topological matter. Band structure is probed through optical excitation from the ground state to Rydberg levels, revealing symmetry-protected topological edge states at zero energy. Edge-state energies are robust to perturbations of tunneling-rates that preserve chiral symmetry, but can be shifted by the introduction of on-site potentials. 

In this paper we present a reconstruction technique for the reduction of unsteady flow data based on neural representations of time‐varying vector fields. Our approach is motivated by the large amount of data typically generated in numerical simulations, and in turn the types of data that domain scientists can generatein situthat are compact, yet useful, for post hoc analysis. One type of data commonly acquired during simulation are samples of the flow map, where a single sample is the result of integrating the underlying vector field for a specified time duration. In our work, we treat a collection of flow map samples for a single dataset as a meaningful, compact, and yet incomplete, representation of unsteady flow, and our central objective is to find a representation that enables us to best recover arbitrary flow map samples. To this end, we introduce a technique for learning implicit neural representations of time‐varying vector fields that are specifically optimized to reproduce flow map samples sparsely covering the spatiotemporal domain of the data. We show that, despite aggressive data reduction, our optimization problem — learning a function‐space neural network to reproduce flow map samples under a fixed integration scheme — leads to representations that demonstrate strong generalization, both in the field itself, and using the field to approximate the flow map. Through quantitative and qualitative analysis across different datasets we show that our approach is an improvement across a variety of data reduction methods, and across a variety of measures ranging from improved vector fields, flow maps, and features derived from the flow map.

 

Crustal seismic velocity models provide essential information for many applications including earthquake source properties, simulations of ground motion and related derivative products. We present a systematic workflow for assessing the accuracy of velocity models with full-waveform simulations. The framework is applied to four regional seismic velocity models for southern California: CVM-H15.11, CVM-S4.26, CVM-S4.26.M01 that includes a shallow geotechnical layer, and the model of Berg et al. For each model, we perform 3-D viscoelastic wave propagation simulations for 48 virtual seismic noise sources (down to 2 s) and 44 moderate-magnitude earthquakes (down to 2 s generally and 0.5 s for some cases) assuming a minimum shear wave velocity of 200 m s–1. The synthetic waveforms are compared with observations associated with both earthquake records and noise cross-correlation data sets. We measure, at multiple period bands for well-isolated seismic phases, traveltime delays and normalized zero-lag cross-correlation coefficients between the synthetic and observed data. The obtained measurements are summarized using the mean absolute derivation of time delay and the mean correlation coefficient. These two metrics provide reliable statistical representations of model quality with consistent results in all data sets. In addition to assessing the overall (average) performance of different models in the entire study area, we examine spatial variations of the models’ quality. All examined models show good phase and waveform agreements for surface waves at periods longer than 5 s, and discrepancies at shorter periods reflecting small-scale heterogeneities and near-surface structures. The model performing best overall is CVM-S4.26.M01. The largest misfits for both body and surface waves are in basin structures and around large fault zones. Inaccuracies generated in these areas may affect tomography and model simulation results at other regions. The seismic velocity models for southern California can be improved by adding better resolved structural representations of the shallow crust and volumes around the main faults.

 

Shellular Funicular Structures (SFSs) are single-layer, two-manifold structures with anticlastic curvature, designed in the context of graphic statics. They are considered as efficient structures applicable to many functions on different scales. Due to their complex geometry, design and fabrication of SFSs are quite challenging, limiting their application in large scales. Furthermore, designing these structures for a predefined boundary condition, control, and manipulation of their geometry are not easy tasks. Moreover, fabricating these geometries is mostly possible using additive manufacturing techniques, requiring a lot of support in the printing process. Cellular funicular structures (CFSs) as strut-based spatial structures can be easily designed and manipulated in the context of graphic statics. This paper introduces a computational algorithm for translating a Cellular Funicular Structure (CFS) to a Shellular Funicular Structure (SFS). Furthermore, it explains a fabrication method to build the structure out of a flat sheet of material using the origami/ kirigami technique as an ideal choice because of its accessibility, processibility, low cost, and applicability to large scales. The paper concludes by displaying a design and fabricated structure using this technique. 

Space frames are widely used in spatial constructions as they are lightweight, rigid, and efficient. However, when it comes to the complex and irregular spaces frames, they can be difficult to fabricate because of the uniqueness of the nodes and bars. This paper presents a novel timber space frame system that can be easily manufactured using 3-axis CNC machines, and therefore increase the ease of the design and construction of complex space frames. The form-finding of the space frame is achieved with the help of polyhedral graphic statics (PGS), and resulted form has inherent planarity which can be harnessed in the materialization of the structure. Inspired by the traditional wood tectonics kerf bending and zippers are applied when devising the connection details. This system's design approach and computational process are described, and a test fabrication of a single node is made via 3-axis CNC milling and both physically and numerically tested. The structural performance shows its potential for applications in large-scale spatial structures.