arXiv subjects

Abhinav Kumar

Publications and source records attributed to Abhinav Kumar.

At least 91 records · Page 5Linked to original sources

arXiv2021

The application of artificial intelligence in software engineering: a review challenging conventional wisdom

The field of artificial intelligence (AI) is witnessing a recent upsurge in research, tools development, and deployment of applications. Multiple software companies are shifting their focus to developing intelligent systems; and many others are deploying AI paradigms to their existing processes. In parallel, the academic research community is injecting AI paradigms to provide solutions to traditional engineering problems. Similarly, AI has evidently been proved useful to software engineering (SE). When one observes the SE phases (requirements, design, development, testing, release, and maintenance), it becomes clear that multiple AI paradigms (such as neural networks, machine learning, knowledge-based systems, natural language processing) could be applied to improve the process and eliminate many of the major challenges that the SE field has been facing. This survey chapter is a review of the most commonplace methods of AI applied to SE. The review covers methods between years 1975-2017, for the requirements phase, 46 major AI-driven methods are found, 19 for design, 15 for development, 68 for testing, and 15 for release and maintenance. Furthermore, the purpose of this chapter is threefold; firstly, to answer the following questions: is there sufficient intelligence in the SE lifecycle? What does applying AI to SE entail? Secondly, to measure, formulize, and evaluate the overlap of SE phases and AI disciplines. Lastly, this chapter aims to provide serious questions to challenging the current conventional wisdom (i.e., status quo) of the state-of-the-art, craft a call for action, and to redefine the path forward.

cs.SE↗

arXiv2021

The history and future prospects of open data and open source software

Open data for all New Yorkers is the tagline on New York City's open data website. Open government is being promoted at most countries of the western world. Government transparency levels are being measured by the amount of data they share through their online public repositories. Additionally, open source software is promoted at governments, academia, and the industry. This is the new digital story of this century, and the new testament between the Gods of technology and there users. Data and software openness will redefine the path forward and aim to rekindle our collective intelligence. Data and software openness can redefine Data Democracy and be the catalyst for its progress. This chapter provides a historical insight into data and software openness, the beginnings, the heroes, prospects for the future, and all things we cannot afford to negotiate or lose.

cs.SI↗

arXiv2021

Misinformation Detection on YouTube Using Video Captions

Millions of people use platforms such as YouTube, Facebook, Twitter, and other mass media. Due to the accessibility of these platforms, they are often used to establish a narrative, conduct propaganda, and disseminate misinformation. This work proposes an approach that uses state-of-the-art NLP techniques to extract features from video captions (subtitles). To evaluate our approach, we utilize a publicly accessible and labeled dataset for classifying videos as misinformation or not. The motivation behind exploring video captions stems from our analysis of videos metadata. Attributes such as the number of views, likes, dislikes, and comments are ineffective as videos are hard to differentiate using this information. Using caption dataset, the proposed models can classify videos among three classes (Misinformation, Debunking Misinformation, and Neutral) with 0.85 to 0.90 F1-score. To emphasize the relevance of the misinformation class, we re-formulate our classification problem as a two-class classification - Misinformation vs. others (Debunking Misinformation and Neutral). In our experiments, the proposed models can classify videos with 0.92 to 0.95 F1-score and 0.78 to 0.90 AUC ROC.

cs.LG↗

arXiv2021

User Pairing and Power Allocation for IRS-Assisted NOMA Systems with Imperfect Phase Compensation

In this letter, we analyze the performance of the intelligent reflecting surface (IRS) assisted downlink non-orthogonal multiple access (NOMA) systems in the presence of imperfect phase compensation. We derive an upper bound on the imperfect phase compensation to achieve minimum required data rates for each user. Using this bound, we propose an adaptive user pairing algorithm to maximize the network throughput. We then derive bounds on the power allocation factors and propose power allocation algorithms for the paired users to achieve the maximum sum rate or ensure fairness. Through extensive simulations, we show that the proposed algorithms significantly outperform the state-of-the-art algorithms.

cs.IT↗

arXiv2021

Non-orthogonal Multiple Access for Multi-cell Indoor VLC

In this letter, we propose a non-orthogonal multiple access (NOMA) based scheme for multi-cell indoor visible light communications (VLC), where cell-edge user is jointly served by multiple cells. Unlike the typical designs in the existing literature, the proposed scheme doesn't involve complex domain and direct current bias addition, which makes the design simple and feasible for real time implementation in indoor VLC. The symbol error rate (SER) of the proposed NOMA scheme is analysed using analytical and simulation results, and is compared with orthogonal multiple access (OMA). It is observed that the average SER of the users with the proposed NOMA is marginally degraded as compared to OMA. However, the SER of cell-edge user with the proposed NOMA is significantly improved with joint maximum likelihood decoding, and outperforms successive interference cancellation based decoding and OMA, with trade-off on computational complexity.

cs.IT↗

arXiv2021

Rate-Splitting Random Access Mechanism for Massive Machine Type Communications in 5G Cellular Internet-of-Things

The cellular Internet-of-Things has resulted in the deployment of millions of machine-type communication (MTC) devices. These massive number of devices must communicate with a single gNodeB (gNB) via the random access channel (RACH) mechanism. However, existing RACH mechanisms are inefficient when dealing with such large number of devices. To address this issue, we propose the rate-splitting random access (RSRA) mechanism, which uses rate splitting and decoding in rate-splitting multiple access (RSMA) to improve RACH success rates. The proposed mechanism divides the message into common and private messages and enhances the decoding performance. We demonstrate, using extensive simulations, that the proposed RSRA mechanism significantly improves the success rate of MTC in cellular IoT networks. We also evaluate the performance of the proposed mechanism with increasing number of devices and received power difference.

cs.IT↗

arXiv2021

GrooMeD-NMS: Grouped Mathematically Differentiable NMS for Monocular 3D Object Detection

Modern 3D object detectors have immensely benefited from the end-to-end learning idea. However, most of them use a post-processing algorithm called Non-Maximal Suppression (NMS) only during inference. While there were attempts to include NMS in the training pipeline for tasks such as 2D object detection, they have been less widely adopted due to a non-mathematical expression of the NMS. In this paper, we present and integrate GrooMeD-NMS -- a novel Grouped Mathematically Differentiable NMS for monocular 3D object detection, such that the network is trained end-to-end with a loss on the boxes after NMS. We first formulate NMS as a matrix operation and then group and mask the boxes in an unsupervised manner to obtain a simple closed-form expression of the NMS. GrooMeD-NMS addresses the mismatch between training and inference pipelines and, therefore, forces the network to select the best 3D box in a differentiable manner. As a result, GrooMeD-NMS achieves state-of-the-art monocular 3D object detection results on the KITTI benchmark dataset performing comparably to monocular video-based methods. Code and models at https://github.com/abhi1kumar/groomed_nms

cs.CV↗

arXiv2020

Adaptive User Pairing for Downlink NOMA System with Imperfect SIC

Non-orthogonal multiple access (NOMA) has been recognized as a key driving technology for the fifth generation (5G) and beyond 5G cellular networks. For a practical dowlink NOMA system with imperfect successive interference cancellation (SIC), we derive bounds on channel coefficients and power allocation factors between NOMA users to achieve higher rates than an equivalent orthogonal multiple access (OMA) system. We propose an adaptive user pairing (A-UP) algorithm for NOMA systems. Through extensive simulations, we show that NOMA with imperfect SIC is not always superior to OMA. Further, the proposed A-UP algorithm results in better performance than state-of-the-art NOMA pairing algorithms in presence of SIC imperfections.

eess.SP↗

arXiv2020

NOMA for Multiple Access Channel and Broadcast Channel in Indoor VLC

Orthogonal frequency division multiplexing (OFDM) based non-orthogonal multiple access (NOMA) has increased complexity and reduced spectral efficiency in visible light communications (VLC) NOMA compared to radiofrequency NOMA due to non-negative real value constraints on transmit symbols. To address this issue, we propose a generalized non-OFDM based scheme for two scenarios of indoor VLC; i) Multiple access channel (MAC), ii) Broadcast channel (BC). We evaluate the performance of the proposed scheme for MAC using successive interference cancellation (SIC) based decoding, joint maximum likelihood (JML) decoding, and a combination of SIC and JML decoding. For BC, we evaluate the performance using SIC based decoding. It is observed that, for MAC, the proposed scheme with JML decoding performs better than the state-of-the-art orthogonal multiple access both in terms of bit error rate (BER) and computations. For BC, the proposed scheme is computationally efficient with trade-off on BER.

eess.SP↗

arXiv2020

LUVLi Face Alignment: Estimating Landmarks' Location, Uncertainty, and Visibility Likelihood

Modern face alignment methods have become quite accurate at predicting the locations of facial landmarks, but they do not typically estimate the uncertainty of their predicted locations nor predict whether landmarks are visible. In this paper, we present a novel framework for jointly predicting landmark locations, associated uncertainties of these predicted locations, and landmark visibilities. We model these as mixed random variables and estimate them using a deep network trained with our proposed Location, Uncertainty, and Visibility Likelihood (LUVLi) loss. In addition, we release an entirely new labeling of a large face alignment dataset with over 19,000 face images in a full range of head poses. Each face is manually labeled with the ground-truth locations of 68 landmarks, with the additional information of whether each landmark is unoccluded, self-occluded (due to extreme head poses), or externally occluded. Not only does our joint estimation yield accurate estimates of the uncertainty of predicted landmark locations, but it also yields state-of-the-art estimates for the landmark locations themselves on multiple standard face alignment datasets. Our method's estimates of the uncertainty of predicted landmark locations could be used to automatically identify input images on which face alignment fails, which can be critical for downstream tasks.

cs.CV↗

arXiv2020

Lossless Compression of Deep Neural Networks

Deep neural networks have been successful in many predictive modeling tasks, such as image and language recognition, where large neural networks are often used to obtain good accuracy. Consequently, it is challenging to deploy these networks under limited computational resources, such as in mobile devices. In this work, we introduce an algorithm that removes units and layers of a neural network while not changing the output that is produced, which thus implies a lossless compression. This algorithm, which we denote as LEO (Lossless Expressiveness Optimization), relies on Mixed-Integer Linear Programming (MILP) to identify Rectified Linear Units (ReLUs) with linear behavior over the input domain. By using L1 regularization to induce such behavior, we can benefit from training over a larger architecture than we would later use in the environment where the trained neural network is deployed.

cs.LG↗

arXiv2019

A Probabilistic Approach to Model SIC based RACH mechanism for Massive Machine Type Communications in Cellular Networks

In a cellular Internet of Things, burst transmissions from millions of machine type communications (MTC) devices can result in channel congestion. The main bottleneck in such scenario is inefficient random access channel (RACH) mechanism that is used to attach MTC devices to a base station (BS). To address this issue of congestion in RACH mechanism, 3GPP has proposed an extended access barring (3GPP-EAB) mechanism. However, several works indicate that the performance of the 3GPP-EAB mechanism can be further improved. In this work, a successive interference cancellation (SIC) based RACH mechanism is considered to significantly increase the success rate and reduce congestion. In the proposed mechanism, the devices are allowed to transmit repeatedly for a finite number of times in a given cycle, and thereafter, the success rate is improved by applying back-and-forth SIC at the BS. A novel probabilistic approach of the proposed mechanism is presented with all transition and steady-state probabilities. Further, the probability of SIC for a given slot is derived. Through extensive numerical results, it is shown that the proposed mechanism significantly outperforms the existing ones in terms of the success rate. Moreover, to obtain the maximum success rate, the optimum number of devices to be entered in a cycle is also calculated.

cs.NI↗

arXiv2019

Equivalent and Approximate Transformations of Deep Neural Networks

Two networks are equivalent if they produce the same output for any given input. In this paper, we study the possibility of transforming a deep neural network to another network with a different number of units or layers, which can be either equivalent, a local exact approximation, or a global linear approximation of the original network. On the practical side, we show that certain rectified linear units (ReLUs) can be safely removed from a network if they are always active or inactive for any valid input. If we only need an equivalent network for a smaller domain, then more units can be removed and some layers collapsed. On the theoretical side, we constructively show that for any feed-forward ReLU network, there exists a global linear approximation to a 2-hidden-layer shallow network with a fixed number of units. This result is a balance between the increasing number of units for arbitrary approximation with a single layer and the known upper bound of $\lceil log(n_0+1)\rceil +1$ layers for exact representation, where $n_0$ is the input dimension. While the transformed network may require an exponential number of units to capture the activation patterns of the original network, we show that it can be made substantially smaller by only accounting for the patterns that define linear regions. Based on experiments with ReLU networks on the MNIST dataset, we found that $l_1$-regularization and adversarial training reduces the number of linear regions significantly as the number of stable units increases due to weight sparsity. Therefore, we can also intentionally train ReLU networks to allow for effective loss-less compression and approximation.

cs.LG↗

arXiv2019

Higgs-Axion interplay and anomalous magnetic phase diagram in TlCuCl$_3$

What is so unique in TlCuCl3 which drives so many unique magnetic features in this compound? To study these properties, here we employ a combination of ab-initio band structure, tight-binding model, and an effective quantum field theory. Within a density-functional theory (DFT) calculation, we find an unexpected bulk Dirac cone without spin-orbit coupling (SOC). Tracing back to its origin, we identify, for the first time, the presence of a Su-Schrieffer-Heeger (SSH) like dimerized Cu chain lying in the 3D crystal structure. The SSH chain, combined with SOC, stipulates an anisotropic 3D Dirac cone where chiral and helical states are intertwined. As a Heisenberg interaction is introduced, we show that the dimerized Cu sublattices of the SSH chain condensate into spin-singlet, dimerized magnets. In the magnetic ground state, we also find a topological phase, distinguished by the axion angle. Finally, to study how the topological axion term couples to magnetic excitations, we derive a Chern-Simons-Ginzburg-Landau action from the 3D SSH Hamiltonian. We find that axion term provides an additional mass term to the Higgs mode, and a lifetime to paramagnons, which are independent of the quantum critical physics. The axion-Higgs interplay can be probed with electric and magnetic field applied parallel or anti-parallel to each other.

cond-mat.str-el↗

arXiv2019

MIMO Codes for Uniform Illumination across Space and Time in VLC with Dimming Control

Indoor visible light communications (VLC) require simultaneous illumination and communication. Hence, uniformity in the illumination is a key consideration for user comfort and data transfer in VLC systems. Several run-length limited codes have been proposed in the literature which mitigate flicker for single input and single output VLC systems. Recently, codes have been proposed for multiple input multiple output (MIMO) VLC systems. However, uniform illumination along with dimming control has not been considered for MIMO VLC systems. Hence, in this paper, we propose codes with generalized algorithms for encoding and decoding that maintain consistency in the illumination across space and time and achieve the desired dimming level. We present the expressions for code rate, run-length, and Hamming distance for the proposed codes. Through Monte Carlo simulations, we compare the codeword error rate (CER) performance of the proposed codes with zero-forcing, minimum mean square error, and maximum likelihood detectors for various number of transmit and receive antennas. We also compare the mutual information of the proposed codes for various number of transmit and receive antennas. The numerical results show that the proposed codes exhibit good performance while satisfying the design criteria.

eess.SP↗

arXiv2019

Location reference identification from tweets during emergencies: A deep learning approach

Twitter is recently being used during crises to communicate with officials and provide rescue and relief operation in real time. The geographical location information of the event, as well as users, are vitally important in such scenarios. The identification of geographic location is one of the challenging tasks as the location information fields, such as user location and place name of tweets are not reliable. The extraction of location information from tweet text is difficult as it contains a lot of non-standard English, grammatical errors, spelling mistakes, non-standard abbreviations, and so on. This research aims to extract location words used in the tweet using a Convolutional Neural Network (CNN) based model. We achieved the exact matching score of 0.929, Hamming loss of 0.002, and $F_1$-score of 0.96 for the tweets related to the earthquake. Our model was able to extract even three- to four-word long location references which is also evident from the exact matching score of over 92\%. The findings of this paper can help in early event localization, emergency situations, real-time road traffic management, localized advertisement, and in various location-based services.

cs.LG↗

arXiv2018

Prediction of low energy phase transition in metal doped MoTe$_2$ from first principle calculations

Metal-insulator transitions in two dimensional materials represent a great opportunity for fast, low energy and ultra-dense switching devices. Due to the small energy difference between its semimetallic and semiconducting crystal phases, phase transition in MoTe$_2$ can occur with an unprecedented small amount of external perturbations. In this work, we used density functional theory to predict critical strain and electrostatic voltage required to control the phase transition of 3d and 4d metal doped MoTe$_2$. We found that small doping contents dramatically affect the relative energies of MoTe$_2$ crystal phases and can largely reduced the energy input to trigger the transition, compared to pristine case. Moreover, the kinetics corresponding to the phase transition in the proposed doped materials are several order of magnitude faster than in MoTe$_2$. For example, we predict 6.3 \% Mn doped MoTe$_2$ to switch phase under 1.19 V gate voltage in less than 1 $μ$s with an input energy of 0.048 aJ/nm$^3$. Due to the presence of dopant, the controlled change of phase is often complemented with a change in magnetic moment leading to multi-functional phase transition.

cond-mat.mtrl-sci↗

arXiv2018

Parametric Synthesis of Text on Stylized Backgrounds using PGGANs

We describe a novel method of generating high-resolution real-world images of text where the style and textual content of the images are described parametrically. Our method combines text to image retrieval techniques with progressive growing of Generative Adversarial Networks (PGGANs) to achieve conditional generation of photo-realistic images that reflect specific styles, as well as artifacts seen in real-world images. We demonstrate our method in the context of automotive license plates. We assess the impact of varying the number of training images of each style on the fidelity of the generated style, and demonstrate the quality of the generated images using license plate recognition systems.

cs.CV↗