Deeksha Shravani - 2021 Poster Contest Resources

Browse: Abstracts, Winners and runners up, Awards Ceremony (Watch Recording / View Slides), Posters by HPCC Systems InternsPosters by Academic Partners, Poster Judges, Virtual Judging, Home

Deeksha Shravani is studying for a Bachelor of Computer Science and Engineering at the RV College of Engineering, Bengaluru, India.

Deeksha is currently researching Virtual Reality (VR) and the utilisation of Big Data platforms for large scale computing. HPCC Systems provides this capability in addition to machine learning functionalities, making it a comprehensive platform for the development of integrated applications. Her interests include computer science, machine learning, cloud technologies and their applications.

Poster Abstract

This talk introduces a Virtual Reality (VR) based online shopping platform and its integration with a recommendation system with the demonstration of the virtual environment. With the advent of the pandemic, the ability of virtual reality platforms to provide a realistic shopping experience puts it in a unique position that assures safety and isolation while also offering the benefits of online shopping platforms to both customers and retailers.

To foster user adoption and improve the experience of the user beyond the confines of traditional shopping experiences, a recommendation system is necessary in such a platform. For a recommendation system in a retail context, the amount of training data present is very vast and warrants the evaluation of the various platforms and Big Data Analytics frameworks which facilitate training of large-scale data.

For a dataset with 9M+ records, the comparison of the training on a Graphics Processing Unit (GPU), HPCC Systems and Spark on Hadoop is performed and various metrics are evaluated. The metrics of evaluation are not only with respect to the performance of the system but include metrics such as time for complete training, time for initial epoch completion etc. which are required for the operational sustainability of the platform. Moreover, deriving conclusions in real time is essential for any recommendation system. In addition to the training time, the time taken for generating inferences i.e., the time required to provide recommendations after training, also is examined. The results indicate the advantages of big data platforms over GPU training. Though Spark on Hadoop is faster in training, the results feature HPCC Systems as the better platform for real time inferencing.

Presentation

In this Video Recording, Deeksha provides a tour and explanation of her poster content.

Developing a Recommendation System for a Virtual Reality based Supermarket using Big Data Platforms

Click on the poster for a larger image. 

All pages in this wiki are subject to our site usage guidelines.