arXiv subjects

Joseph Wang

Publications and source records attributed to Joseph Wang.

At least 19 recordsLinked to original sources

arXiv2022

Wakeword Detection under Distribution Shifts

We propose a novel approach for semi-supervised learning (SSL) designed to overcome distribution shifts between training and real-world data arising in the keyword spotting (KWS) task. Shifts from training data distribution are a key challenge for real-world KWS tasks: when a new model is deployed on device, the gating of the accepted data undergoes a shift in distribution, making the problem of timely updates via subsequent deployments hard. Despite the shift, we assume that the marginal distributions on labels do not change. We utilize a modified teacher/student training framework, where labeled training data is augmented with unlabeled data. Note that the teacher does not have access to the new distribution as well. To train effectively with a mix of human and teacher labeled data, we develop a teacher labeling strategy based on confidence heuristics to reduce entropy on the label distribution from the teacher model; the data is then sampled to match the marginal distribution on the labels. Large scale experimental results show that a convolutional neural network (CNN) trained on far-field audio, and evaluated on far-field audio drawn from a different distribution, obtains a 14.3% relative improvement in false discovery rate (FDR) at equal false reject rate (FRR), while yielding a 5% improvement in FDR under no distribution shift. Under a more severe distribution shift from far-field to near-field audio with a smaller fully connected network (FCN) our approach achieves a 52% relative improvement in FDR at equal FRR, while yielding a 20% relative improvement in FDR on the original distribution.

cs.SD↗

arXiv2022

Latency Control for Keyword Spotting

Conversational agents commonly utilize keyword spotting (KWS) to initiate voice interaction with the user. For user experience and privacy considerations, existing approaches to KWS largely focus on accuracy, which can often come at the expense of introduced latency. To address this tradeoff, we propose a novel approach to control KWS model latency and which generalizes to any loss function without explicit knowledge of the keyword endpoint. Through a single, tunable hyperparameter, our approach enables one to balance detection latency and accuracy for the targeted application. Empirically, we show that our approach gives superior performance under latency constraints when compared to existing methods. Namely, we make a substantial 25\% relative false accepts improvement for a fixed latency target when compared to the baseline state-of-the-art. We also show that when our approach is used in conjunction with a max-pooling loss, we are able to improve relative false accepts by 25 % at a fixed latency when compared to cross entropy loss.

eess.AS↗

arXiv2021

Investigating Degradation Modes in Zn-AgO Aqueous Batteries with $\textit{In-Situ}$ X-ray Micro Computed Tomography

To meet growing energy demands, degradation mechanisms of energy storage devices must be better understood. As a non-destructive tool, X-ray Computed Tomography (CT) has been increasingly used by the battery community to perform $\textit{in-situ}$ experiments that can investigate dynamic phenomena. However, few have used X-ray CT to study representative battery systems over long cycle lifetimes (>100 cycles). Here, we report the $\textit{in-situ}$ CT study of Zn-Ag batteries and demonstrate the effects of current collector parasitic gassing over long-term storage and cycling. We design performance representative $\textit{in-situ}$ CT cells that can achieve >250 cycles at a high areal capacity of $\mathrm{12.5\;mAh/cm^2}$. Combined with electrochemical experiments, the effects of current collector parasitic gassing are revealed with micro-scale CT (MicroCT). The volume expansion and evolution of ZnO and Zn depletion is quantified with cycling and elevated temperature testing. The experimental insights are then utilized to develop larger form-factor $\mathrm{4\;cm^2}$ cells with electrochemically compatible current collectors. With this, we demonstrate over 325 cycles at a high capacity of $\mathrm{12.5\;mAh/cm^2}$ for a $\mathrm{4\;cm^2}$ form-factor. This work demonstrates that $\textit{in-situ}$ X-ray CT used in long cycle-lifetime studies can be applied to examine a multitude of other battery chemistries to improve their performances.

cond-mat.mtrl-sci↗

arXiv2020

Self-Healing Small-Scale Swimmers

Herein, self-healing small-scale swimmers capable of autonomous propulsion and on-the-fly structural recovery are described. The new strategy instantaneously restores the functionality of the swimmer after it has suffered severe damage. Incorporation of magnetic microparticles in strips along with the printed functional body layers (consisting of conductive carbon, low-density hydrophobic polymer and catalytically active metal) results in rapid reorientation and reattachment of the moving damaged catalytic tail to its complimentary broken static body piece. Such magnetic alignment and attraction restores the original swimmer structure and propulsion behavior, independent of user input, displaying healing efficiencies as high as 88 percent. Modeling of the magnetic fields and simulations of various swimmer configurations are used to study the magnetic force field distribution around the swimmers. The influence of the damage position and pattern of multiple magnetic healing strips is examined and their influence upon healing efficiency is compared. Owing to its versatility, fast recovery, simplicity, and efficiency, the new on-the-fly self-healing strategy represents an important step towards the development of new classes of robots that can regain their functionality in situations of extreme mechanical damage where repair is not possible or challenging.

cond-mat.soft↗

arXiv2020

High Performance Printed AgO-Zn Rechargeable Battery for Flexible Electronics

The rise of flexible electronics calls for cost-effective and scalable batteries with good mechanical and electrochemical performance. In this work, we developed printable, polymer-based AgO-Zn batteries that feature flexibility, rechargeability, high areal capacity, and low impedance. Using elastomeric substrate and binders, the current collectors, electrodes, and separators can be easily screen-printed layer-by-layer and vacuum-sealed in a stacked configuration. The batteries are customizable in sizes and capacities, with the highest obtained areal capacity of 54 mAh/cm2 for primary applications. Advanced micro-CT and EIS were used to characterize the battery, whose mechanical stability was tested with repeated twisting and bending. The batteries were used to power a flexible E-ink display system that requires a high-current drain and exhibited superior performance than commercial coin-cell batteries. The developed battery presents a practical solution for powering a wide range of electronics and holds major implications for the future development of practical and high-performance flexible batteries.

physics.app-ph↗

arXiv2018

Probabilistic Semantic Retrieval for Surveillance Videos with Activity Graphs

We present a novel framework for finding complex activities matching user-described queries in cluttered surveillance videos. The wide diversity of queries coupled with unavailability of annotated activity data limits our ability to train activity models. To bridge the semantic gap we propose to let users describe an activity as a semantic graph with object attributes and inter-object relationships associated with nodes and edges, respectively. We learn node/edge-level visual predictors during training and, at test-time, propose to retrieve activity by identifying likely locations that match the semantic graph. We formulate a novel CRF based probabilistic activity localization objective that accounts for mis-detections, mis-classifications and track-losses, and outputs a likelihood score for a candidate grounded location of the query in the video. We seek groundings that maximize overall precision and recall. To handle the combinatorial search over all high-probability groundings, we propose a highest precision subgraph matching algorithm. Our method outperforms existing retrieval methods on benchmarked datasets.

cs.MM↗

arXiv2017

Adaptive Neural Networks for Efficient Inference

We present an approach to adaptively utilize deep neural networks in order to reduce the evaluation time on new examples without loss of accuracy. Rather than attempting to redesign or approximate existing networks, we propose two schemes that adaptively utilize networks. We first pose an adaptive network evaluation scheme, where we learn a system to adaptively choose the components of a deep network to be evaluated for each example. By allowing examples correctly classified using early layers of the system to exit, we avoid the computational time associated with full evaluation of the network. We extend this to learn a network selection system that adaptively selects the network to be evaluated for each example. We show that computational time can be dramatically reduced by exploiting the fact that many examples can be correctly classified using relatively efficient networks and that complex, computationally costly networks are only necessary for a small fraction of examples. We pose a global objective for learning an adaptive early exit or network selection policy and solve it by reducing the policy learning problem to a layer-by-layer weighted binary classification problem. Empirically, these approaches yield dramatic reductions in computational cost, with up to a 2.8x speedup on state-of-the-art networks from the ImageNet image recognition challenge with minimal (<1%) loss of top5 accuracy.

cs.LG↗

arXiv2017

Field of Groves: An Energy-Efficient Random Forest

Machine Learning (ML) algorithms, like Convolutional Neural Networks (CNN), Support Vector Machines (SVM), etc. have become widespread and can achieve high statistical performance. However their accuracy decreases significantly in energy-constrained mobile and embedded systems space, where all computations need to be completed under a tight energy budget. In this work, we present a field of groves (FoG) implementation of random forests (RF) that achieves an accuracy comparable to CNNs and SVMs under tight energy budgets. Evaluation of the FoG shows that at comparable accuracy it consumes ~1.48x, ~24x, ~2.5x, and ~34.7x lower energy per classification compared to conventional RF, SVM_RBF , MLP, and CNN, respectively. FoG is ~6.5x less energy efficient than SVM_LR, but achieves 18% higher accuracy on average across all considered datasets.

cs.DC↗

arXiv2016

Pruning Random Forests for Prediction on a Budget

We propose to prune a random forest (RF) for resource-constrained prediction. We first construct a RF and then prune it to optimize expected feature cost & accuracy. We pose pruning RFs as a novel 0-1 integer program with linear constraints that encourages feature re-use. We establish total unimodularity of the constraint set to prove that the corresponding LP relaxation solves the original integer program. We then exploit connections to combinatorial optimization and develop an efficient primal-dual algorithm, scalable to large datasets. In contrast to our bottom-up approach, which benefits from good RF initialization, conventional methods are top-down acquiring features based on their utility value and is generally intractable, requiring heuristics. Empirically, our pruning algorithm outperforms existing state-of-the-art resource-constrained algorithms.

stat.ML↗

arXiv2016

Resource Constrained Structured Prediction

We study the problem of structured prediction under test-time budget constraints. We propose a novel approach applicable to a wide range of structured prediction problems in computer vision and natural language processing. Our approach seeks to adaptively generate computationally costly features during test-time in order to reduce the computational cost of prediction while maintaining prediction performance. We show that training the adaptive feature generation system can be reduced to a series of structured learning problems, resulting in efficient training using existing structured learning algorithms. This framework provides theoretical justification for several existing heuristic approaches found in literature. We evaluate our proposed adaptive system on two structured prediction tasks, optical character recognition (OCR) and dependency parsing and show strong performance in reduction of the feature costs without degrading accuracy.

stat.ML↗

arXiv2016

Optimally Pruning Decision Tree Ensembles With Feature Cost

We consider the problem of learning decision rules for prediction with feature budget constraint. In particular, we are interested in pruning an ensemble of decision trees to reduce expected feature cost while maintaining high prediction accuracy for any test example. We propose a novel 0-1 integer program formulation for ensemble pruning. Our pruning formulation is general - it takes any ensemble of decision trees as input. By explicitly accounting for feature-sharing across trees together with accuracy/cost trade-off, our method is able to significantly reduce feature cost by pruning subtrees that introduce more loss in terms of feature cost than benefit in terms of prediction accuracy gain. Theoretically, we prove that a linear programming relaxation produces the exact solution of the original integer program. This allows us to use efficient convex optimization tools to obtain an optimally pruned ensemble for any given budget. Empirically, we see that our pruning algorithm significantly improves the performance of the state of the art ensemble method BudgetRF.

stat.ML↗

arXiv2015

Efficient Learning by Directed Acyclic Graph For Resource Constrained Prediction

We study the problem of reducing test-time acquisition costs in classification systems. Our goal is to learn decision rules that adaptively select sensors for each example as necessary to make a confident prediction. We model our system as a directed acyclic graph (DAG) where internal nodes correspond to sensor subsets and decision functions at each node choose whether to acquire a new sensor or classify using the available measurements. This problem can be naturally posed as an empirical risk minimization over training data. Rather than jointly optimizing such a highly coupled and non-convex problem over all decision nodes, we propose an efficient algorithm motivated by dynamic programming. We learn node policies in the DAG by reducing the global objective to a series of cost sensitive learning problems. Our approach is computationally efficient and has proven guarantees of convergence to the optimal system for a fixed architecture. In addition, we present an extension to map other budgeted learning problems with large number of sensors to our DAG architecture and demonstrate empirical performance exceeding state-of-the-art algorithms for data composed of both few and many sensors.

stat.ML↗

arXiv2015

Sensor Selection by Linear Programming

We learn sensor trees from training data to minimize sensor acquisition costs during test time. Our system adaptively selects sensors at each stage if necessary to make a confident classification. We pose the problem as empirical risk minimization over the choice of trees and node decision rules. We decompose the problem, which is known to be intractable, into combinatorial (tree structures) and continuous parts (node decision rules) and propose to solve them separately. Using training data we greedily solve for the combinatorial tree structures and for the continuous part, which is a non-convex multilinear objective function, we derive convex surrogate loss functions that are piecewise linear. The resulting problem can be cast as a linear program and has the advantage of guaranteed convergence, global optimality, repeatability and computational efficiency. We show that our proposed approach outperforms the state-of-art on a number of benchmark datasets.

stat.ML↗

arXiv2015

Feature-Budgeted Random Forest

We seek decision rules for prediction-time cost reduction, where complete data is available for training, but during prediction-time, each feature can only be acquired for an additional cost. We propose a novel random forest algorithm to minimize prediction error for a user-specified {\it average} feature acquisition budget. While random forests yield strong generalization performance, they do not explicitly account for feature costs and furthermore require low correlation among trees, which amplifies costs. Our random forest grows trees with low acquisition cost and high strength based on greedy minimax cost-weighted-impurity splits. Theoretically, we establish near-optimal acquisition cost guarantees for our algorithm. Empirically, on a number of benchmark datasets we demonstrate superior accuracy-cost curves against state-of-the-art prediction-time algorithms.

stat.ML↗

arXiv2015

Max-Cost Discrete Function Evaluation Problem under a Budget

We propose novel methods for max-cost Discrete Function Evaluation Problem (DFEP) under budget constraints. We are motivated by applications such as clinical diagnosis where a patient is subjected to a sequence of (possibly expensive) tests before a decision is made. Our goal is to develop strategies for minimizing max-costs. The problem is known to be NP hard and greedy methods based on specialized impurity functions have been proposed. We develop a broad class of \emph{admissible} impurity functions that admit monomials, classes of polynomials, and hinge-loss functions that allow for flexible impurity design with provably optimal approximation bounds. This flexibility is important for datasets when max-cost can be overly sensitive to "outliers." Outliers bias max-cost to a few examples that require a large number of tests for classification. We design admissible functions that allow for accuracy-cost trade-off and result in $O(\log n)$ guarantees of the optimal cost among trees with corresponding classification accuracy levels.

cs.LG↗

arXiv2011

Structural Similarity and Distance in Learning

We propose a novel method of introducing structure into existing machine learning techniques by developing structure-based similarity and distance measures. To learn structural information, low-dimensional structure of the data is captured by solving a non-linear, low-rank representation problem. We show that this low-rank representation can be kernelized, has a closed-form solution, allows for separation of independent manifolds, and is robust to noise. From this representation, similarity between observations based on non-linear structure is computed and can be incorporated into existing feature transformations, dimensionality reduction techniques, and machine learning methods. Experimental results on both synthetic and real data sets show performance improvements for clustering, and anomaly detection through the use of structural similarity.

stat.ML↗

arXiv2011

Biomolecular Filters for Improved Separation of Output Signals in Enzyme Logic Systems Applied to Biomedical Analysis

Biomolecular logic systems processing biochemical input signals and producing "digital" outputs in the form of YES/NO were developed for analysis of physiological conditions characteristic of liver injury, soft tissue injury and abdominal trauma. Injury biomarkers were used as input signals for activating the logic systems. Their normal physiological concentrations were defined as logic-0 level, while their pathologically elevated concentrations were defined as logic-1 values. Since the input concentrations applied as logic 0 and 1 values were not sufficiently different, the output signals being at low and high values (0, 1 outputs) were separated with a short gap making their discrimination difficult. Coupled enzymatic reactions functioning as a biomolecular signal processing system with a built-in filter property were developed. The filter process involves a partial back-conversion of the optical-output-signal-yielding product, but only at its low concentrations, thus allowing the proper discrimination between 0 and 1 output values.

cond-mat.soft↗

arXiv2011

High-Speed Propulsion of Flexible Nanowire Motors: Theory and Experiments

Micro/nano-scale propulsion has attracted considerable recent attention due to its promise for biomedical applications such as targeted drug delivery. In this paper, we report on a new experimental design and theoretical modelling of high-speed fuel-free magnetically-driven propellers which exploit the flexibility of nanowires for propulsion. These readily prepared nanomotors display both high dimensional propulsion velocities (up to ~ 21 micrometer per second) and dimensionless speeds (in body lengths per revolution) when compared with natural microorganisms and other artificial propellers. Their propulsion characteristics are studied theoretically using an elastohydrodynamic model which takes into account the elasticity of the nanowire and its hydrodynamic interaction with the fluid medium. The critical role of flexibility in this mode of propulsion is illustrated by simple physical arguments, and is quantitatively investigated with the help of an asymptotic analysis for small-amplitude swimming. The theoretical predictions are then compared with experimental measurements and we obtain good agreement. Finally, we demonstrate the operation of these nanomotors in a real biological environment (human serum), emphasizing the robustness of their propulsion performance and their promise for biomedical applications.

physics.flu-dyn↗