Yash Desai's repositories
Aconex-Notes
Uploading my persoonal obsidian notes for some examinations
Aspect-Based-Sentiment-Analysis-
Given a review and an aspect, we classify the sentiment conveyed towards that aspect on a three-point scale: POSITIVE, NEUTRAL, and NEGATIVE.
Aspect-Based-Sentiment-Analysis-with-BERT
Using a pre-trained BERT (Bidirectional Embedding Representations from Transformers) model as a trainable keras layer
Audio-Classification---Classifying-Audio-Files-into-8-songs-from-MLEnd-Hums-and-Whistles-Dataset
Classifying the hums and whistles from MLEnd Hums and Whistles dataset into 8 different songs. Formulated 2 solutions (basic and advanced). Basic solution involves binary classification while the advanced solution involves more complex techniques of feature extraction and multi-class classification.
Badges4-README.md-Profile
:octocat: Improve your README.md profile with these amazing badges.
Comparing-Text-Classification-Models-
Evaluating the performance of various text classification models and experimenting with various types of word embeddings and neural Networks layers.
Fashion-MNIST-Classification-using-Deep-Learning
Creating a Deep Learning model which classifies images from the Fashion MNIST Dataset to 10 classes.
Neural-Machine-Translation
A neural machine translation model based on the sequence-to-sequence (seq2seq) models proposed by Sutskever et al., 2014 and Cho et al., 2014. The seq2seq model is widely used in machine translation systems such as Google’s neural machine translation system (GNMT) (Wu et al., 2016).
Speech-Emotion-Recognition
Recognises Emotion
Coreference-Resolution
Building a coreference system based on the mention-ranking algorithm proposed by Lee et al (2017)
CRF-Tagging-in-Movie-Queries
Classifying sequences from the movie queries dataset into 23 different CRF tags
Data-Mining
Data Mining techniques demonstrates as assignments during my Masters at the Queen Mary University of London
Dataset_Creation
Creating a dataset for MSc Project
Dialogue-Act-Tagging
Here we'll look at two different DA classification models. The Switchboard Dialog Act Corpus is being used for training.
Download-Previews-and-Extract-Features
Code used for downloading song previews from spotify and extracting audio features from the same.
Ethereum-Analysis
Demonstrated the use of Map-Reduce and PySpark to compute average transactions per month, top smart contracts and top active miners using Ethereum's transaction and block data from Google BigQuery. Further, performed scam analysis and fork analysis to find the most lucrative type of scam and change in price after the fork.
ivy
The Unified Machine Learning Framework
Leet-Code-Solutions
Adding my solutions for LeetCode problems while I try to become a better programmer and prepare for job interviews
Modalysis
Cloud Computing Mini Project
Named-Entity-Resolver-
Training a Named Entity Resolver using bidirectional GRU and Multi-layer FFNN
personality-prediction
Predicting personality and moral values of people using music listening history of people.
Skip-gram-Model-for-Word2Vec
Training a skip-gram neural network model to obtain word embeddings.
Text-Classification-using-LSTM
Building an LSTM model to classify movie reviews as either positive or negative
Vector-Space-Semantics-for-Similary-between-Eastenders-Characters
Creating a vector representation of a document containing lines spoken by a character in the Eastenders script data. Further computing cosine similarity and improving the same using pre-processinf techniques and adding dialogue context.
wasm-ref
A web assembly file reference for a custom power bi dashboard
yash-td
Config files for my GitHub profile.