arXiv subjects

Piyush Gupta

Publications and source records attributed to Piyush Gupta.

At least 37 records · Page 2Linked to original sources

arXiv2020

Incentivizing Collaboration in Heterogeneous Teams via Common-Pool Resource Games

We consider a team of heterogeneous agents that is collectively responsible for servicing, and subsequently reviewing, a stream of homogeneous tasks. Each agent has an associated mean service time and a mean review time for servicing and reviewing the tasks, respectively. Agents receive a reward based on their service and review admission rates. The team objective is to collaboratively maximize the number of "serviced and reviewed" tasks. We formulate a Common-Pool Resource (CPR) game and design utility functions to incentivize collaboration among heterogeneous agents in a decentralized manner. We show the existence of a unique Pure Nash Equilibrium (PNE), and establish convergence of best response dynamics to this unique PNE. Finally, we establish an analytic upper bound on three measures of inefficiency of the PNE, namely the price of anarchy, the ratio of the total review admission rate, and the ratio of latency, along with an empirical study.

math.OC↗

arXiv2020

MixBoost: Synthetic Oversampling with Boosted Mixup for Handling Extreme Imbalance

Training a classification model on a dataset where the instances of one class outnumber those of the other class is a challenging problem. Such imbalanced datasets are standard in real-world situations such as fraud detection, medical diagnosis, and computational advertising. We propose an iterative data augmentation method, MixBoost, which intelligently selects (Boost) and then combines (Mix) instances from the majority and minority classes to generate synthetic hybrid instances that have characteristics of both classes. We evaluate MixBoost on 20 benchmark datasets, show that it outperforms existing approaches, and test its efficacy through significance testing. We also present ablation studies to analyze the impact of the different components of MixBoost.

cs.LG↗

arXiv2020

Retrospective Loss: Looking Back to Improve Training of Deep Neural Networks

Deep neural networks (DNNs) are powerful learning machines that have enabled breakthroughs in several domains. In this work, we introduce a new retrospective loss to improve the training of deep neural network models by utilizing the prior experience available in past model states during training. Minimizing the retrospective loss, along with the task-specific loss, pushes the parameter state at the current training step towards the optimal parameter state while pulling it away from the parameter state at a previous training step. Although a simple idea, we analyze the method as well as to conduct comprehensive sets of experiments across domains - images, speech, text, and graphs - to show that the proposed loss results in improved performance across input domains, tasks, and architectures.

cs.CV↗

arXiv2020

Explain Your Move: Understanding Agent Actions Using Specific and Relevant Feature Attribution

As deep reinforcement learning (RL) is applied to more tasks, there is a need to visualize and understand the behavior of learned agents. Saliency maps explain agent behavior by highlighting the features of the input state that are most relevant for the agent in taking an action. Existing perturbation-based approaches to compute saliency often highlight regions of the input that are not relevant to the action taken by the agent. Our proposed approach, SARFA (Specific and Relevant Feature Attribution), generates more focused saliency maps by balancing two aspects (specificity and relevance) that capture different desiderata of saliency. The first captures the impact of perturbation on the relative expected reward of the action to be explained. The second downweighs irrelevant features that alter the relative expected rewards of actions other than the action to be explained. We compare SARFA with existing approaches on agents trained to play board games (Chess and Go) and Atari games (Breakout, Pong and Space Invaders). We show through illustrative examples (Chess, Atari, Go), human studies (Chess), and automated evaluation methods (Chess) that SARFA generates saliency maps that are more interpretable for humans than existing approaches. For the code release and demo videos, see https://nikaashpuri.github.io/sarfa-saliency/.

cs.CV↗

arXiv2020

Towards Safer Self-Driving Through Great PAIN (Physically Adversarial Intelligent Networks)

Automated vehicles' neural networks suffer from overfit, poor generalizability, and untrained edge cases due to limited data availability. Researchers synthesize randomized edge-case scenarios to assist in the training process, though simulation introduces potential for overfit to latent rules and features. Automating worst-case scenario generation could yield informative data for improving self driving. To this end, we introduce a "Physically Adversarial Intelligent Network" (PAIN), wherein self-driving vehicles interact aggressively in the CARLA simulation environment. We train two agents, a protagonist and an adversary, using dueling double deep Q networks (DDDQNs) with prioritized experience replay. The coupled networks alternately seek-to-collide and to avoid collisions such that the "defensive" avoidance algorithm increases the mean-time-to-failure and distance traveled under non-hostile operating conditions. The trained protagonist becomes more resilient to environmental uncertainty and less prone to corner case failures resulting in collisions than the agent trained without an adversary.

cs.LG↗

arXiv2020

ShapeVis: High-dimensional Data Visualization at Scale

We present ShapeVis, a scalable visualization technique for point cloud data inspired from topological data analysis. Our method captures the underlying geometric and topological structure of the data in a compressed graphical representation. Much success has been reported by the data visualization technique Mapper, that discreetly approximates the Reeb graph of a filter function on the data. However, when using standard dimensionality reduction algorithms as the filter function, Mapper suffers from considerable computational cost. This makes it difficult to scale to high-dimensional data. Our proposed technique relies on finding a subset of points called landmarks along the data manifold to construct a weighted witness-graph over it. This graph captures the structural characteristics of the point cloud, and its weights are determined using a Finite Markov Chain. We further compress this graph by applying induced maps from standard community detection algorithms. Using techniques borrowed from manifold tearing, we prune and reinstate edges in the induced graph based on their modularity to summarize the shape of data. We empirically demonstrate how our technique captures the structural characteristics of real and synthetic data sets. Further, we compare our approach with Mapper using various filter functions like t-SNE, UMAP, LargeVis and show that our algorithm scales to millions of data points while preserving the quality of data visualization.

cs.LG↗

arXiv2019

OpticalGAN : Generative Adversarial Networks for Continuous Variable Quantum Computation

We present OpticalGAN, an extension of quantum generative adversarial networks for continuous-variable quantum computation. OpticalGAN consists of photonic variational circuits comprising of optical Gaussian and Kerr gates. Photonic quantum computation is a realization of continuous variable quantum computing which involves encoding and processing information in the continuous quadrature amplitudes of quantized electromagnetic field such as light. Information processing in photonic quantum computers is performed using optical gates on squeezed light. Both the generator and discriminator of OpticalGAN are short depth variational circuits composed of gaussian and non-gaussian gates. We demonstrate our approach by using OpticalGAN to generate energy eigenstates and coherent states. All of our code is available at https://github.com/abcd1729/opticalgan.

cs.ET↗

arXiv2019

Multimapper: Data Density Sensitive Topological Visualization

Mapper is an algorithm that summarizes the topological information contained in a dataset and provides an insightful visualization. It takes as input a point cloud which is possibly high-dimensional, a filter function on it and an open cover on the range of the function. It returns the nerve simplicial complex of the pullback of the cover. Mapper can be considered a discrete approximation of the topological construct called Reeb space, as analysed in the $1$-dimensional case by [Carriere et al.,2018]. Despite its success in obtaining insights in various fields such as in [Kamruzzaman et al., 2016], Mapper is an ad hoc technique requiring lots of parameter tuning. There is also no measure to quantify goodness of the resulting visualization, which often deviates from the Reeb space in practice. In this paper, we introduce a new cover selection scheme for data that reduces the obscuration of topological information at both the computation and visualisation steps. To achieve this, we replace global scale selection of cover with a scale selection scheme sensitive to local density of data points. We also propose a method to detect some deviations in Mapper from Reeb space via computation of persistence features on the Mapper graph.

math.AT↗

arXiv2018

MAGIX: Model Agnostic Globally Interpretable Explanations

Explaining the behavior of a black box machine learning model at the instance level is useful for building trust. However, it is also important to understand how the model behaves globally. Such an understanding provides insight into both the data on which the model was trained and the patterns that it learned. We present here an approach that learns if-then rules to globally explain the behavior of black box machine learning models that have been used to solve classification problems. The approach works by first extracting conditions that were important at the instance level and then evolving rules through a genetic algorithm with an appropriate fitness function. Collectively, these rules represent the patterns followed by the model for decisioning and are useful for understanding its behavior. We demonstrate the validity and usefulness of the approach by interpreting black box models created using publicly available data sets as well as a private digital marketing data set.

cs.AI↗

arXiv2015

Energy-Efficient Communication in the Presence of Synchronization Errors

Communication systems are traditionally designed to have tight transmitter-receiver synchronization. This requirement has negligible overhead in the high-SNR regime. However, in many applications, such as wireless sensor networks, communication needs to happen primarily in the energy-efficient regime of low SNR, where requiring tight synchronization can be highly suboptimal. In this paper, we model the noisy channel with synchronization errors as an insertion/deletion/substitution channel. For this channel, we propose a new communication scheme that requires only loose transmitter-receiver synchronization. We show that the proposed scheme is asymptotically optimal for the Gaussian channel with synchronization errors in terms of energy efficiency as measured by the rate per unit energy. In the process, we also establish that the lack of synchronization causes negligible loss in energy efficiency. We further show that, for a general discrete memoryless channel with synchronization errors and a general input cost function admitting a zero-cost symbol, the rate per unit cost achieved by the proposed scheme is within a factor two of the information-theoretic optimum.

cs.IT↗

arXiv2015

On the Scaling of Interference Alignment Under Delay and Power Constraints

Future wireless standards such as 5G envision dense wireless networks with large number of simultaneously connected devices. In this context, interference management becomes critical in achieving high spectral efficiency. Orthogonal signaling, which limits the number of users utilizing the resource simultaneously, gives a sum-rate that remains constant with increasing number of users. An alternative approach called interference alignment promises a throughput that scales linearly with the number of users. However, this approach requires very high SNR or long time duration for sufficient channel variation, and therefore may not be feasible in real wireless systems. We explore ways to manage interference in large networks with delay and power constraints. Specifically, we devise an interference phase alignment strategy that combines precoding and scheduling without using power control to exploit the diversity inherent in a system with large number of users. We show that this scheme achieves a sum-rate that scales almost logarithmically with the number of users. We also show that no scheme using single symbol phase alignment, which is asymmetric complex signaling restricted to a single complex symbol, can achieve better than logarithmic scaling of the sum-rate.

cs.IT↗

arXiv2014

Energy-Efficient Communication over the Unsynchronized Gaussian Diamond Network

Communication networks are often designed and analyzed assuming tight synchronization among nodes. However, in applications that require communication in the energy-efficient regime of low signal-to-noise ratios, establishing tight synchronization among nodes in the network can result in a significant energy overhead. Motivated by a recent result showing that near-optimal energy efficiency can be achieved over the AWGN channel without requiring tight synchronization, we consider the question of whether the potential gains of cooperative communication can be achieved in the absence of synchronization. We focus on the symmetric Gaussian diamond network and establish that cooperative-communication gains are indeed feasible even with unsynchronized nodes. More precisely, we show that the capacity per unit energy of the unsynchronized symmetric Gaussian diamond network is within a constant factor of the capacity per unit energy of the corresponding synchronized network. To this end, we propose a distributed relaying scheme that does not require tight synchronization but nevertheless achieves most of the energy gains of coherent combining.

cs.IT↗

arXiv2012

Towards a Queueing-Based Framework for In-Network Function Computation

We seek to develop network algorithms for function computation in sensor networks. Specifically, we want dynamic joint aggregation, routing, and scheduling algorithms that have analytically provable performance benefits due to in-network computation as compared to simple data forwarding. To this end, we define a class of functions, the Fully-Multiplexible functions, which includes several functions such as parity, MAX, and k th -order statistics. For such functions we exactly characterize the maximum achievable refresh rate of the network in terms of an underlying graph primitive, the min-mincut. In acyclic wireline networks, we show that the maximum refresh rate is achievable by a simple algorithm that is dynamic, distributed, and only dependent on local information. In the case of wireless networks, we provide a MaxWeight-like algorithm with dynamic flow splitting, which is shown to be throughput-optimal.

cs.NI↗

arXiv2012

Bounds on Minimum Number of Anchors for Iterative Localization and its Connections to Bootstrap Percolation

Iterated localization is considered where each node of a network needs to get localized (find its location on 2-D plane), when initially only a subset of nodes have their location information. The iterated localization process proceeds as follows. Starting with a subset of nodes that have their location information, possibly using global positioning system (GPS) devices, any other node gets localized if it has three or more localized nodes in its radio range. The newly localized nodes are included in the subset of nodes that have their location information for the next iteration. This process is allowed to continue, until no new node can be localized. The problem is to find the minimum size of the initially localized subset to start with so that the whole network is localized with high probability. There are intimate connections between iterated localization and bootstrap percolation, that is well studied in statistical physics. Using results known in bootstrap percolation, we find a sufficient condition on the size of the initially localized subset that guarantees the localization of all nodes in the network with high probability.

cs.IT↗

arXiv2011

Distributed Resource Allocation for Proportional Fairness in Multi-Band Wireless Systems

A challenging problem in multi-band multi-cell self-organized wireless systems, such as multi-channel Wi-Fi networks, femto/pico cells in 3G/4G cellular networks, and more recent wireless networks over TV white spaces, is of distributed resource allocation. This involves four components: channel selection, client association, channel access, and client scheduling. In this paper, we present a unified framework for jointly addressing the four components with the global system objective of maximizing the clients throughput in a proportionally fair manner. Our formulation allows a natural dissociation of the problem into two sub-parts. We show that the first part, involving channel access and client scheduling, is convex and derive a distributed adaptation procedure for achieving Pareto-optimal solution. For the second part, involving channel selection and client association, we develop a Gibbs-sampler based approach for local adaptation to achieve the global objective, as well as derive fast greedy algorithms from it that achieve good solutions.

cs.NI↗

arXiv2010

When is a Function Securely Computable?

A subset of a set of terminals that observe correlated signals seek to compute a given function of the signals using public communication. It is required that the value of the function be kept secret from an eavesdropper with access to the communication. We show that the function is securely computable if and only if its entropy is less than the "aided secret key" capacity of an associated secrecy generation model, for which a single-letter characterization is provided.

cs.IT↗

arXiv2010

A Decentralized Approach for Service Discovery & Availability in P-Grids

The widespread emergence of the Internet as a platform for electronic data distribution and the advent of structured information have revolutionized our ability to deliver information to any corner of the world. Although Service Oriented Architecture (SOA) is a paradigm for organizing and utilizing distributed capabilities that may be under the control of different ownership domains and implemented using various technology stacks and every organization may not be geared up for this. To harness the various software / service resources placed on various systems, we have proposed and implemented a model that is able to establish discovery and sharing in load balanced P-grid environment. The experimental results show that the proposed approach has dramatically lowered the network traffic (nearly negligible), while achieving load balancing in P2P grid systems. Our model is able to support discovery and sharing of resources also.

cs.NI↗

arXiv2010

The Balanced Unicast and Multicast Capacity Regions of Large Wireless Networks

We consider the question of determining the scaling of the $n^2$-dimensional balanced unicast and the $n 2^n$-dimensional balanced multicast capacity regions of a wireless network with $n$ nodes placed uniformly at random in a square region of area $n$ and communicating over Gaussian fading channels. We identify this scaling of both the balanced unicast and multicast capacity regions in terms of $Θ(n)$, out of $2^n$ total possible, cuts. These cuts only depend on the geometry of the locations of the source nodes and their destination nodes and the traffic demands between them, and thus can be readily evaluated. Our results are constructive and provide optimal (in the scaling sense) communication schemes.

cs.IT↗