INTRODUCTION
My Research focus is on “High Performance Machine Learning” with emphasis on developing algorithms to analyze real time IoT sensors data and implement them on parallel and distributed systems. Working on high performance Machine Learning requires deep understanding of Machine Learning algorithms and detailed knowledge of parallel systems on which they are going to be applied. I have acquired these skills from my academic and industrial research careers. With my diverse skills and experiences, I plan to develop technology for addressing challenges in creating cost effective IoT enabled sensors for collecting useful information, Big Data Analysis of collected information, and Parallel Computing to bring technology to life.

PAST AND CURRENT RESEARCH
A. Machine Learning Algorithms for Phylogenetic Analysis (Bioinformatics)

My research during PhD had been on developing computationally efficient machine learning algorithms for analyzing and finding relationships between species. The study of relationship among species or genes with the combination of molecular biology and mathematics in known as Phylogenetics. My contributions were in genomic data collection of different species, developing customized machine learning techniques to analyze it. I developed and implemented cluster ensemble methods for making phylogenetic trees, combining different trees from different sources and research groups to make a tree of life.

During few experiments we found the phylogeny don’t always takes form of a tree, especially in viruses. After discussing with experts in phylogenetic studies we developed classification methods with overlapping classes to make phylogenetic networks. Later we also used ancestral divergence time and labeling to improve accuracy of the classification.

B. Machine Learning Algorithm Application in Medical Imaging (Industrial Experience)

Over the past seven years of my industrial career, I have gained expertise in application of machine learning methods with medical image reconstruction and implementing them on parallel computing architectures with thousands of cores to achieve best performance in market. To achieve best performance I focused on developing efficient data structures, adopting machine learning algorithms, efficient utilization of parallel resources such as computing cores and memory. I also gave an extensive effort towards quantitative evaluation and comparison of the final product.

During this research, I developed solutions to Parallel Imaging and Combining data from multiple
Interpolate the missing information using different machine learning algorithms. I developed and implemented a solution on parallel computing architecture, to reconstruct the image.

Another interesting work is related to combining similar information acquired by different sources into one. Due to the advancement in RF technology, these days MR data is acquired by large set of independent receive channels. This help in improving SNR and Parallel Imaging performance. However, processing the data received by multiple channels leads to increased memory and computational loads of reconstruction. I developed mathematical framework to reduce the redundancy and applied it using GPU to produce high SNR images in an acceptable time.

FUTURE RESEARCH PLAN
A. Internet of Things (IoT) and Cloud Computing Research
During last couple of years internet started taking a different direction with the advancements in research and production of IoT based chips that are highly power efficient and cloud computing capable. Now it looks possible to gather huge amounts of information using sensors, store and process it on cloud to extract pattern that are useful for big business houses as well as an individuals. When power efficient and communication ready chips are integrated with different sensors, they will change the way we take daily decisions and have a big impact on the life.
These advancements in sensor and communication technology come with greater challenges focusing on information filtering, storing data, and processing huge amounts of data collected from millions of sensors in real time. Every IoT application will have different challenges to be addressed and new algorithms to be proposed. For example, pulse sensors on a patient need to address the challenges of real time data transfer, storage and analysis. Whereas, a sensor deployed to detect available parking spot needs only an event based data transfer and storage, and when sufficient data is available, it can be used for different behavioral studies to improve quality of services.
I would like to develop sensors and big data analysis resources for supporting transportation, agriculture, mechanical industries (such as oil industry) and patient monitoring systems. The main goal of my research will be to address methods to generate useful date and patterns that can be used millions of users across the globe to improve their quality of life.

B. Machine Learning Research
In past, I have been solving problems based on machine learning where data was acquired using only on source or similar source. We also know important features in acquired data. In future, When we have multiple cameras for monitoring systems connected through internet and transmitting images and videos and text information to the could we will immediately face challenges in finding right information to be transmitted over the communication to reduce the traffic over communication network, storing and analyzing data. I would like to work on real time feature selection algorithms to address this issue. I also would like to explore event based data processing approaches for data selection and transmission.
When it comes to monitoring systems we acquire and transmit different data for analysis. For example a patients monitoring system data will have images and video data, pulse, temperature, heart rate monitoring sensors data in different format. We need to develop advance methods to analyze them together to help us finding right patterns accurately.

SERVICE TO UNIVERSITY
I wish to use my industrial and academic experience in Machine Learning, Parallel Programming and Software Engineering to contribute to the ongoing and potential research in Machine Learning and Big Data on IoT sensor data. I am willing to collaborate with other faculties and researchers to achieve the research goals and expectations of university. I wish to serve the department and university in translational research by bridging the gap between industry needs and academic goals.

great inspiration for the celebration of new year -2015

pls make video for Phd also