Shloka Shah

Software Development Engineer at HackerRank
Computer Science Undergraduate | B.Tech at Sardar Patel Institute Of Technology | 2021
Winter Intern'21 at HackerRank | GHCI'20 Scholar | SIH'20 Finalist
Mumbai, India
Deep Learning | ML Enthusiast | Web Development | Data Science | AWS

Publications



1. MeghNA: Cloud Detection and Motion Prediction in INSAT Images

In this paper, we propose a set of algorithms for cloud detection and nowcasting using INSAT satellite imagery.

                 Paper Link: https://link.springer.com/chapter/10.1007/978-981-16-4369-9_11

Key Terms: Conv-LSTMS, Deep learning, Satellite imagery, Forecasting, Mask RCNN, Cloud motion, Cloud detection


2. Segro: Key Towards Modern Waste Management.

This paper focuses on the fact that with modernization and advancement in technology, we can improve the overall urban waste management system. SEGRO is an end to end solution, encompassing waste disposal, segregation, collection and the effective usage of recycled waste. We propose a solution that will not only effectively segregate the waste into different types based on their life-cycle but also will provide an optimal algorithm for the collection of waste.

                 Paper Link: https://ieeexplore.ieee.org/abstract/document/9154113

Key Terms: waste segregation, route optimization, recycling waste, waste classifier, smart bins, deep learning


3. Spaced Repetition for Slow Learners

An algorithm ‘Spaced Repetition for Slow Learners’ (SRSL) is described to schedule repetitions which eventually adapts to the capacity of the learner. SRSL computes the score of learners for a particular assessment based on factors such as response time, difficulty and dependency of questions. The exponential forgetting curve model is the memory model assumed by SRSL. Based on this algorithm, a model has been proposed with experimental analysis of the same.

                 Paper Link: https://ieeexplore.ieee.org/document/9332189

Key Terms: Spaced Repetition, Slow learners, Forgetting curve, Exponential decay, Retention capacity


4. Prosodic Speech Synthesis of Narratives Depicting Emotional Diversity Using Deep Learning

In this paper, an Expressive Text-to-Speech Synthesis System (ETSSS) is proposed which considers the dominant emotions in the text provided. ETSSS works in two parts: first, it identifies the label behind the text, and second produces expressive speech.

                 Paper Link: https://link.springer.com/chapter/10.1007/978-981-16-4369-9_4

Key Terms: Speech synthesis, Expressive speech, Emotional speech, Emotion recognition, BERT, Deep learning, Convolutional neural networks