I'm a Machine Learning Enthusiast, an Independent Researcher and Technology Explorer.
My areas of interest are Deep Learning , Machine Learning and Natural User Interfaces (NUI).
I am exploring how various DL algorithms can be incorporated into various use cases and scenarios to solve real-time problems, by automating the process and creating intelligent solutions.
Apres makes it simple for companies to develop AI in regulated industries to explain decisions in a traceable way. This reduces data management over 10X while enabling a fully transparent environment for AI development. We use your training data to create an explainable version of your AI model’s behaviour. Apres uses this model to explain historic, present and future decisions, critical to developing transparent AI.
Building Deep Learning solutions on scale for enterprises. Major areas of focus are Generative Artificial Intelligence, Continues learning platform using Reinforcement Learning.
Building NLU engines to support Nordic languages like Danish, Italian, etc. Major areas of focus are Conversational Agents, Designing CNN and RNN architectures for complex use cases and Reinforcement Learning for active learning.
Building Cognitive solutions around chatbots, Q/A engines for mCaaS. Focusing on implementing NLP engines using Statistical and Probabilistic Modelling.
I explore and implement machine learning algorithms for various use cases - NLP, Feature extraction, Classification, Predictive model, etc. My current areas of R&D focus include NLP parsing, speech to text conversion and Q/A service. Previously, I have also explored Biometrics for POC's and client project delivery.
Rohit Khokher, R.C. Singh, Rahul Kumar,
Palmprint Recognition Using Geometrical and Texture Properties,
2nd International Conference on Research in Science, Engineering and Technology (ICRSET 2014), DUBAI (UAE),
pp: 53-57, ISBN: 978-93-82242-81-9
Rohit Khokher, R.C. Singh, Rahul Kumar,
Footprint Recognition with Principal Component Analysis and Independent Component Analysis,
International Conference on Science and Engineering of Materials (ICSEM 2014),
DOI: 10.1002/masy.201400045
Rohit Khokher, R.C. Singh, Saiyam Diwan, Rahul Kumar,
Generation of private 512-Bit Security using Human Palm Print,
International Conference on Recent technologies in Advanced Electronics (ICRTAE 2013),
pp: 176-193 ,
ISBN: 978-3-659-27725-2
Code Contributions
Rohit Khokher, R.C. Singh
Footprint-Based Personal Recognition Using Dactyloscopy Technique,
Industrial Mathematics and Complex Systems, October 2017
pp: 207-219, Online ISBN: 978-981-10-3758-0
Rohit Khokher, R.C. Singh
Footprint-Based Personal Recognition using Scanning Technique,
Indian Journal of Science and Technology, Vol 9(44), November 2016
Online ISBN: 0974-5645
Created a Docker container to deploy the Tensorflow models in production.Implemented a text classification model using Convolutional Neural Netowork and used web.py to make a RESTful service.
Created a Danish word2vec model using Tensorflow with the embedding_size = 300 and vocab_size = ~324k. Used TensorBoard for embeddings visualization. Feel free to check it out at link
Developed a NLU framework incorporating various Deep Learning architecture to handel NLP tasks. This f/w currently supports Danish, Italian and English languages. All the intelligence in chatbots developed by BotSupply is power by Pachamama.
Developed and packaged as RESTfull service multi-class classifier using Tensorflow. Implemented 3 layers of Convolutional Neural Network for feature engineering with 97.2% of accuracy.
Developed a python utility to clean the data by removing the special unicode characters from the csv file and deployed it as a web service. Working on its user interface in AngularJS.
Integrated Amazon Alexa with RaspberryPi 2 and developed a software trigger to automate the voice command initiation process. Looking further to automate it with Gesture recognition or Face recognition.
Developed an Android app which performs facial expression recognition and triggers some basic activities such as calling, playing music, etc.
Worked in a team of 5 to create a health care application in Java and C++. Working to make the application more intelligent using Computer Vision and Machine Learning concepts. This application was featured on Prabhat Khabar.
Worked under the guidance of Assis. Prof. Kiran Ravulakollu. Developed a Biometric based security system using human Footprints and Palmprints. Wrote the GUI in Java and backend in Matlab.
Worked under the guidance of Asso. Prof. Rohit Khokher to develop a human identification system using human Footprints.
Developed an algorithm to extract principal line components from the palmprint images using Morphological Operations and optimized the efficiency of the system by ~13 ms using Radon Tranformation.
Worked under the guidance of Late Prof. Pervez Ahmed to code and optimize 4 parameters out of 1700 in human handwriting. This project aims to identify the human characterstics using handwriting pattern.
Implemented a realtime eye detection using OpenCV in ObjectiveC. Used HAAR classifier to detect the frontal face. Also integrated with the imbuild webcam. Demo
