Atul Dhingra

Atul Dhingra

San Francisco Bay Area
12K followers 500+ connections

Join to follow

About

As a Machine Learning Engineer at PayPal, I am driven by a passion for enabling the…

Contributions

Activity

Join now to see all activity

Experience

  • PayPal Graphic

    PayPal

    San Jose, California, United States

  • -

    San Francisco Bay Area

  • -

    San Francisco Bay Area

  • -

    San Jose, California, United States

  • -

    New Jersey, United States

  • -

  • -

  • -

  • -

    New Delhi

  • -

    New Delhi Area, India

  • -

    Netaji Subhas Institute Of Technology, Delhi, India

  • -

    Hyderabad Area, India

  • -

    New Delhi Area, India

  • -

    New Delhi

Education

Licenses & Certifications

Join now to see all certifications

Volunteer Experience

  • Joy of Giving Week

    Rotaract Club of NSIT Regency

    - 1 month

    Poverty Alleviation

    Have been involved in volunteering in various events such as Joy of Giving Week, Collection drives etc. I have also had opportunities to interact with various NGOs like Goonj, Jaagriti, Masoom Duniya and Avanti-Fellows. Towards my senior year i was also responsible for organizing these events apart from volunteering in it. During my final two years with the club, I was involved in managing the finances of the club.

  • Coordinator, Joy of Giving Week

    Rotaract Club of NSIT Regency

    - 1 month

    Poverty Alleviation

    For the 4 years I was attached to the club, I volunteered in various events such as Joy of Giving Week, Collection drives etc. There spawned multitudes of opportunities to interact with various NGOs like Goonj, Jaagriti, Masoom Duniya and Avanti-Fellows, which finally led me to start the Avanti Fellows, NSIT Chapter. Towards my senior year I worked more towards organizing events apart from volunteering in it. During my final two years with the club, I primarily managed the finances of the club…

    For the 4 years I was attached to the club, I volunteered in various events such as Joy of Giving Week, Collection drives etc. There spawned multitudes of opportunities to interact with various NGOs like Goonj, Jaagriti, Masoom Duniya and Avanti-Fellows, which finally led me to start the Avanti Fellows, NSIT Chapter. Towards my senior year I worked more towards organizing events apart from volunteering in it. During my final two years with the club, I primarily managed the finances of the club apart from other volunteering and organizing opportunities.

Publications

  • Indian Electoral Roll Corpus

    Harvard Dataverse

    See publication

  • Instate Corpus

    Harvard Dataverse

    See publication

  • How Does Word2Vec Work ?

    ODH Journal Club Meeting

    See publication

  • Model Complexity-Accuracy Trade-off for a Convolutional Neural Network

    arXiv

    Convolutional Neural Networks(CNN) has had a great success in the recent past, because of the advent of faster GPUs and memory access. CNNs are really powerful as they learn the features from data in layers such that they exhibit the structure of the V-1 features of the human brain. A huge bottleneck, in this case, is that CNNs are very large and have a very high memory footprint, and hence they cannot be employed on devices with limited storage such as mobile phone, IoT etc. In this work, we…

    Convolutional Neural Networks(CNN) has had a great success in the recent past, because of the advent of faster GPUs and memory access. CNNs are really powerful as they learn the features from data in layers such that they exhibit the structure of the V-1 features of the human brain. A huge bottleneck, in this case, is that CNNs are very large and have a very high memory footprint, and hence they cannot be employed on devices with limited storage such as mobile phone, IoT etc. In this work, we study the model complexity versus accuracy trade-off on MNIST dataset, and give a concrete framework for handling such a problem, given the worst case accuracy that a system can tolerate. In our work, we reduce the model complexity by 236 times, and memory footprint by 19.5 times compared to the base model while achieving worst case accuracy threshold.

    See publication

  • Face Identification and Clustering

    arXiv

    In this thesis, we study two problems based on clustering algorithms. In the first
    problem, we study the role of visual attributes using an agglomerative clustering algorithm
    to whittle down the search area where the number of classes is high to improve the
    performance of clustering. We observe that as we add more attributes, the clustering
    performance increases overall. In the second problem, we study the role of clustering
    in aggregating templates in a 1:N open set protocol…

    In this thesis, we study two problems based on clustering algorithms. In the first
    problem, we study the role of visual attributes using an agglomerative clustering algorithm
    to whittle down the search area where the number of classes is high to improve the
    performance of clustering. We observe that as we add more attributes, the clustering
    performance increases overall. In the second problem, we study the role of clustering
    in aggregating templates in a 1:N open set protocol using multi-shot video as a probe.
    We observe that by increasing the number of clusters, the performance increases with
    respect to the baseline and reaches a peak, after which increasing the number of clusters
    causes the performance to degrade. Experiments are conducted using recently introduced
    unconstrained IARPA Janus IJB-A, CS2, and CS3 face recognition datasets.

    See publication

  • Robust Speaker Verification using GFCC based i-vectors

    Springer

    Other authors

  • Wielding Audio-Books for Visually Impaired using Gesture Recognition

    International Journal Of Advanced Research Trends In Engineering And Technology, pp 65-70, 2015

    The paper presents various aspects of hand based gesture recognition. The paper discusses two main modules, colour-based tracking and colour-independent tracking. Various aspects of both these modules are discussed in detail in this paper. In the case of Colour Based Tracking we have used RGB model to segment out hand to track gesture. In the latter module, we use a combination of Mixture of Gaussians Background (MOG) model and convex hull followed by convexity
    defects to segment the hand…

    The paper presents various aspects of hand based gesture recognition. The paper discusses two main modules, colour-based tracking and colour-independent tracking. Various aspects of both these modules are discussed in detail in this paper. In the case of Colour Based Tracking we have used RGB model to segment out hand to track gesture. In the latter module, we use a combination of Mixture of Gaussians Background (MOG) model and convex hull followed by convexity
    defects to segment the hand. K-means is used to track the centre of the region of interest (ROI).

    Other authors
    See publication

  • Biometric Based Personal Authentication Using Eye Movement Tracking

    SEMCCO 2013, Part II, LNCS 8298, pp. 248–256, 2013

    The paper provides an insight into the newly emerging field of Eye Movement Tracking (EMT), spanning across various facets of EMT, from acquisition to authentication. The second most cardinal problem of machine learning after overfitting, i.e. Curse of Dimensionality is dealt with using a novel method of error analysis on EMT based personal authentication through a dimensionality reduction algorithm. We apply both static and dynamic methods for the dimensionality reduction in EMT to achieve…

    The paper provides an insight into the newly emerging field of Eye Movement Tracking (EMT), spanning across various facets of EMT, from acquisition to authentication. The second most cardinal problem of machine learning after overfitting, i.e. Curse of Dimensionality is dealt with using a novel method of error analysis on EMT based personal authentication through a dimensionality reduction algorithm. We apply both static and dynamic methods for the dimensionality reduction in EMT to achieve promising results of personal authentication and compare these results based on speed and accuracy of both the methods. A decision tree classifier is used in two cases (static and dynamic) of EMT for the classification. The novel method presented in this paper is not limited to EMT and it can be emulated for other biometric modalities as well.

    Other authors
    See publication

  • Learning Large Scale Sparse Models

    arXiv:2301.10958

  • Neurorehab: An Interface for Rehabilitation

    arXiv:2301.10957

Join now to see all publications

Patents

  • Systems and methods for performing spatial analytics using spatial data related to a cashier-less shopping store for autonomous checkout

    Filed U.S. Application No.: 63/428,373

    Other inventors

  • Machine learning-based re-identification of shoppers in a cashier-less store for autonomous checkout

    Filed U.S. Application No. 17/988,650

    Other inventors
    See patent

  • Subject-tracking in a cashier-less shopping store for autonomous checkout for improving item and shelf placement and for performing spatial analytics using spatial data and the subject-tracking

    U.S. Application No. 18/522,104

    Other inventors

Courses

  • Artificial Intelligence

    -

  • Brain Computer Interaction

    CS-674

  • Circuits and Systems

    -

  • Computer Aided Design

    -

  • Computer Graphics

    -

  • Computer Networks

    -

  • Image and Video Processing, Duke University

    coursera.org

  • Introduction to Algorithms and Data Structures

    CS-512

  • Introduction to Artificial Intelligence

    CS-520

  • Introduction to Programming

    -

  • Machine Learning, University of Washington

    coursera.org

  • Mathematics 1

    -

  • Mathematics 2

    -

  • Mathematics 3

    -

  • Microprocessors (8085)

    -

  • VHDL

    -

Projects

  • Character Level Language Model

    - Present

    Built a character level language model to generate new dinosaur names with a character level RNN

  • Debiasing Word Vectors

    - Present

    Debiasing Word Vectors by neutralizing and equalizing bias for non-gender specific words using GloVe word vectors

  • Emojify: Making text sentences more expressive with emojis

    Emojify: Making text sentences more expressive with emojis, using word embeddings and LSTM

  • k-ary encoder-decoder

    Developed a LSTM based integer to base-k encoder-decoder with tensorflow

  • Machine Translation

    Neural Machine Translation with an attention model to convert human readable dates into machine readable dates using LSTM

  • Music Generation with LSTMs

    Generated novel jazz solos using LSTM networks in a style representative of a body of performed work in training set

  • Trigger Word Detection

    End to end trigger-word detection model using 1D CNN and LSTM on speech spectogram obtained from synthesized speech data

  • Gradient checking and Regularization for a Deep Neural Network

    Gradient checking and Regularization for a Deep Neural Network

  • Optimization for a Deep Neural Network with Gradient Descent, Momentum and Adam

    Optimization for a Deep Neural Network with Gradient Descent, Momentum and Adam

  • Temperature Level Prediction using Thermometer Images

    Developed a CNN based temperature level detection and prediction model from thermometer images

  • Car detection system for autonomous vehicles

    Implemented YOLO model for car detection system for autonomous vehicles using images from dashboard camera

  • Eclectic-Nets: A suite of most popular CNNs

    Eclectic-Nets: A suite of most popular CNNs with keras and tensorflow that include leNet, ResNet-50 etc.

  • Face identification, verification and happiness classification

    A Deep Learning based face identification, verification and happiness classification using FaceNet architecture with triplet loss

  • Image classification task using ResNet-50

    Implemented ResNet-50 model from scratch for image classification task on hand sign(SIGNS dataset)

  • Neural Style Transfer

    Implemented Neural style transfer algorithm for generating novel artistic images of the Louvre museum based on painting of Monet

  • Scene Clustering

    Clustered scenes into groups based on SIFT features with FLANN based matcher using OpenCV

    See project

  • Indian Electoral Rolls

    - Present

    We built a dataset of nearly all the Indian electors. Our data includes information on first and last name, gender, polling station (constituency, district, and state), father or husband's name, among other such details. We assembled this data by scraping and parsing the electoral rolls

    See project

  • Unified Face Landmark and Gender Recognition

    A deep learning-based face detection, landmark detection, and gender classification pipeline in a unified workflow using dlib and keras(tensorflow backend)

    See project

  • Time Series Analysis: Stock Price Prediction

    Stock price prediction using LSTM and 1D Convoltional Layer implemented in keras with TF backend on daily closing price of S&P 500 data from Jan 2000 to Aug 2016

    See project

  • Model Complexity-Accuracy Trade-off for a Convolutional Neural Network

    Convolutional Neural Networks(CNN) has had a great success in the recent fast, because of the advent of faster GPUs and memory access. CNNs are really powerful as they learn the features from data in layers such that they exhibit the structure of the V-1 features of the human brain. A huge bottleneck in this case is that CNNs are very large and have a very high memory footprint, and hence they cannot be employed on devices with limited storage such as mobile phone, IoT etc. In this work we…

    Convolutional Neural Networks(CNN) has had a great success in the recent fast, because of the advent of faster GPUs and memory access. CNNs are really powerful as they learn the features from data in layers such that they exhibit the structure of the V-1 features of the human brain. A huge bottleneck in this case is that CNNs are very large and have a very high memory footprint, and hence they cannot be employed on devices with limited storage such as mobile phone, IoT etc. In this work we study the model complexity versus accuracy trade-off on MNIST dataset, and give a concrete framework of handling such a problem, given the worst case accuracy that a system can tolerate. In our work we reduce the model complexity by 620 times, and memory footprint by 19.5 times compared to base model, while achieving worst case accuracy threshold.

    See project

  • Player Tracking in American Football

    Detect the wide Receiver in a single shot video clip, where the offensive team is to the left

    See project

  • Ethnicity Classifier

    This classifier uses the sub-strings of size 3 to learn probability associated with each substring such that it belongs to a particular ethnic group using the census 2000 data. When a new string is presented, it is broken down into sub-strings of size 3, and prediction is done on each of the sub-string using the model computed before. The output accuracy of the predictor is 85% even in an unbalanced dataset. The model follows the Naive Bayes assumption of conditional independence

    See project

  • Augmented Reality

    Synthesized clip-art and a 3D mesh onto an image for which the homographies have been calculated using the camera caliberation project.

    See project

  • Camera Caliberation

    Calibrate the camera of a robot vehicle for 3d and 2d camera calibration.

    See project

  • Matchmaking via Lasso

    Lasso is implemented on CIFAR-10 dataset via Co-ordinate Gradient descent algorithm

    See project

  • Compuational Photography: Texture Synthesis and Image Inpainting

    This work involves Texture Synthesis and Image Inpainting. The idea of Texture Synthesis by Non-Parametric Sampling proposed by Efros and Leung, ICCV99, and region filling by Criminisi et al “Region Filling and Object Removal by Exemplar-Based Image Inpainting”, TIP 2004 has been implemented.

    See project

  • Face Recognition Algorithms

    Implemented and evaluated four basic face recognition algorithms: Eigenfaces, Fisherfaces, Support Vector Machine (SVM), and Sparse Representation-based Classification (SRC) on YaleB dataset

    See project

  • Optical Character Recognition

    Designed an optical character recognition (OCR) system for hand-written characters, in a fixed-resolution setting.

    See project

  • Spam Classification with Perceptron and Kernel Perceptron

    Spam Classification on the Web Spam Dataset using Percepton and Kernel Perceptron with Polynomial, Gaussian, Exponential and Laplacian Kernels.

    See project

  • Topic Classification with Naive Bayes

    A Naive Bayes algorithm is implemented on the Reuters dataset for topic classification("earn" or "acq")

    See project

  • Face Identification and Clustering

    -

    In this thesis, we study two problems based on clustering algorithms. In the first problem, we study the role of visual attributes using an agglomerative clustering algorithm to whittle down the search area where the number of classes is high to improve the performance of clustering. We observe that as we add more attributes, the clustering performance increases overall. In the second problem, we study the role of clustering in aggregating templates in a 1:N open set protocol using multi-shot…

    In this thesis, we study two problems based on clustering algorithms. In the first problem, we study the role of visual attributes using an agglomerative clustering algorithm to whittle down the search area where the number of classes is high to improve the performance of clustering. We observe that as we add more attributes, the clustering performance increases overall. In the second problem, we study the role of clustering in aggregating templates in a 1:N open set protocol using multi-shot video as a probe. We observe that by increasing the number of clusters, the performance increases with respect to the baseline and reaches a peak, after which increasing the number of clusters causes the performance to degrade. Experiments are conducted using recently introduced unconstrained IARPA Janus IJB-A, CS2, and CS3 face recognition datasets.

    See project

  • Learning Large Scale Sparse Models,

    -

    In this work, we consider learning sparse models in large scale setting, where the number of samples and the feature dimension can grow as large as millions or billions. Two immediate issues occur under such challenging scenarios: (i) com- putational cost; (ii) memory overhead. In particular, the memory issue precludes a large volume of prior algorithms that are based on batch optimization technique. To remedy the problem, we propose to learn sparse models such as Lasso in online manner where…

    In this work, we consider learning sparse models in large scale setting, where the number of samples and the feature dimension can grow as large as millions or billions. Two immediate issues occur under such challenging scenarios: (i) com- putational cost; (ii) memory overhead. In particular, the memory issue precludes a large volume of prior algorithms that are based on batch optimization technique. To remedy the problem, we propose to learn sparse models such as Lasso in online manner where in each iteration, only one randomly chosen sample is revealed to update a sparse iterate. Thereby, the memory cost is independent of the sample size and gradient evaluation for one sample is efficient. Perhaps amazing, we find that with the same parameter, sparsity promoted by batch methods is not preserved in online fashion. We analyze such interesting phenomenon and illustrate some effective variants including mini-batch methods and a hard thresholding based stochastic gradient algorithm. Extensive experiments are carried out on a public dataset which supports our findings and algorithms.

    Other creators
    See project

  • Route-Finding Algorithms

    -

    Implemented various Route-Finding algorithms as a part of the coursework.

    Other creators

  • Algorithm Snippets

    -

    • Created and presented interactive web based algorithm snippets for Minimum-Spanning Tree algorithm
    • The project was awarded 3 rd best in approximately 25 projects.

    Other creators
    See project

  • Neurorehab: An Interface for Rehabilitation

    -

    • Investigated the role of repetitive activities(robust plasticity of motor cortex) for neurorehabilitation with
    the help of a non-invasive brain computer interface.
    • Built a VR rehabilitation game with Unity and C#, and engineered an algorithm to dynamically scale
    with and treat upper-arm disabilities with an immersive game design.

    Other creators
    See project

  • Text-independent Speaker Recognition

    -

    Developed a robust speaker verification algorithm invariant to noise and multi-channel input using GFCC,
    MFCC and i-vectors.
    • The research was published in Springer, Proceedings of the IEEE International Conference on
    Signal, Networks, Computing, and Systems 2016

    Other creators
    • Jeevan Medikonda
    • Bijaya Ketan PanigrahiMada
    • Madasu Hanmandlu
    See project

  • Biometrics Based Applications

    -

    • Explored various applications of Biometrics, including heuristics and authetication based applications as
    a part of my undergraduate thesis.
    • Studied multi-modal aspect of biometrics and correlated various biometric modalities including hand-veins

    Other creators
    • Smriti Srivastava

  • Gesture Recognition

    -

    • Formulated and published gesture recognition modules for controlling Audio-Books for visually impaired
    with OpenCV
    • Compared color-based and color-independent tracking of hand based gesture

    Other creators

  • Eye Movement Tracking

    -

    • Developed and published a novel method for error analysis for eye movement tracking for biometrics.
    • Published the research in Springer,Proceedings of Swarm, Evolutionary and Memetic Computing
    2013 Part II, LNCS(Springer) 8298, pp. 248-256, 2013

    Other creators

  • Light Detection and Bridge Wave Rectifier

    -

    The project was aimed in implementing a Light Detection Circuit using a Light Dependant Resistor, and a Bridge Wave Rectifier, on a Bread Board. so that it could be extrapolated by the organization in fire and explosion applications at the lab, Centre for Fire, Explosion and Environment Safety.

    Other creators
    See project

  • Temperature Measurement and Recording

    -

    Temperature Measuring and Recording System using PIC 16F877A microcontroller by making use of DS18B20 Temperature Sensor. The system was developed to display and then store temperature readings during the experiment. The project was developed so as the system can be implemented by the organization further for temperature measurements in Explosions And Blasts , using the needed sensors.

    Other creators
    See project

  • Iris Authentication

    -

    Iris Recognition is a rapidly expanding method of biometric authentication that uses pattern recognition techniques on images of irides to uniquely identify an individual. The aim of the project was compiling a robust database of human irides and to compare the results of the acquired images with the previous popular available database such as Chinese Academy of Sciences' Institute of Automation (CASIA). The project became a stepping stone towards my pursuit for field of Computer Vision and its…

    Iris Recognition is a rapidly expanding method of biometric authentication that uses pattern recognition techniques on images of irides to uniquely identify an individual. The aim of the project was compiling a robust database of human irides and to compare the results of the acquired images with the previous popular available database such as Chinese Academy of Sciences' Institute of Automation (CASIA). The project became a stepping stone towards my pursuit for field of Computer Vision and its practical applications. This is the first project i have been involved which is the muse for me for going forward in the filed of Computer Vision.

    Other creators
    See project

Honors & Awards

  • Invited Talk on “Navigating the Complexity of ML Production: Insights and Lessons Learned”

    ACM and IEEE Chicago

  • Invited Talk on “Scaling Up Machine Learning: From Research to Production”

    Texas A&M University, Texas, USA

  • Invited Talk on “Transforming ML Ideas into Market-Ready Products”

    Santa Clara University, CA, USA

  • Senior Member, IEEE

    Institute of Electrical and Electronics Engineers

  • Invited Talk on “Machine Learning, Demystified”

    Arizona State University, AZ, USA

  • Invited Talk on “Productionizing ML Systems”

    Productiv Inc., CA, USA

  • Invited Talk on “Intro to Machine Learning”

    Maharaja Surajmal Institute Of Technology, Delhi, India

  • Invited Talk on “Machine Learning in Production”

    Netaji Subhas University of Technology, Delhi, India

  • Invited Talk on “Scaling Machine Learning in Production”

    Jaypee Institute of Information Technology, Delhi, India

  • Invited Talk on “Scaling Machine Learning in Production”

    Jawaharlal Nehru University, Delhi, India

  • Invited Talk on “Learning to Represent Text with Word2Vec”

    Fat Cat Fab Lab, NY

Languages

  • Hindi

    Native or bilingual proficiency

  • English

    Full professional proficiency

Recommendations received

14 people have recommended Atul

Join now to view

More activity by Atul

View Atul’s full profile

  • See who you know in common
  • Get introduced
  • Contact Atul directly
Join to view full profile

Other similar profiles

Explore more posts

Explore collaborative articles

We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.

Explore More

Others named Atul Dhingra

43 others named Atul Dhingra are on LinkedIn

See others named Atul Dhingra

Add new skills with these courses

See all courses