Web-based AI chatbot builder platform

Developed a Machine-Learning-based chatbot platform with data analytics tools developed with Django.

Features:
1. No-code chatbot, easy setup using Django
2. Multiple training options
3. Built-in data analytics and visualization tools
4. Simple UI for viewing MySQL data
5. Administrator management tools

Car Damage Detection

Implementation of Mask R-CNN on Python 3, Keras, and TensorFlow to detect the area of damage on a car. The model generates bounding boxes and segmentation masks for each instance of car in the image. It's based on Feature Pyramid Network (FPN) and a ResNet50/ResNet101 backbone.Photos of damaged car can be input into model to assess damage.

NLP based Q&A chatbot

Developed a Q&A chatbot that answers question in natural language based on the content of the given passage that is easy to integrate with any existing website. This TensorFlow Natural Language Question Answering model is based on a pre-trained BERT model fine-tuned on Stanford Question Answering Dataset (SQuAD) 2.0 dataset.

Automated time-series forecasting

Prediction vs actual result of the FProphet model

Multi-steps forecast result of the FBProphet model

Built an automated system that will fit time-series dataset into different Statistical and Machine Learning Models for multi-step forecasting and return result from the most accurate model.

Model available in the system:
1. ARIMA (Auto Regressive Integrated Moving Average)
2. Prophet by Facebook
3. HWES (Holt Winter’s Exponential Smoothing)

LSTM (Long Short-Term Memory networks) Models for Time Series Forecasting

Univariate Multi-Step time forecasting with Encoder-Decoder Model

Multiple Input Multi-Step forecasting with Stacked LSTM Model

Developed different LSTM (Long Short-Term Memory networks) models to forecast univariate & multivariate time series dataset. Models are evaluated with 8 metrics.

Type of LSTM model:
1. Vanilla LSTM
2. Stacked LSTM
3. Bidirectional LSTM
4. CNN LSTM
5. ConvLSTM

Type of forecasting:
1. Univariate time series forecasting
2. Multivariate time series forecasting
3. Multi-step time series forecasting
4. Multivariate Multi-Step time series forecasting

Deep learning model for Time Series Forecasting

Application of the Temporal Fusion Transformer (TFT), a novel attention-based architecture which combines high-performance multi-horizon forecasting with interpretable insights into temporal dynamics to forecast time series data.

Neural NILM: Deep Neural Networks Applied to Energy Disaggregation

Undergraduate research by Yuzhe Lim in Spring 2019 on the topic of Deep Neural Networks application on NILM (Nonintrusive load monitoring) for Energy Disaggregation.

Two-Class Boosted Decision Tree Model using Azure Machine Learning Studio

Model training pipeline

Model deployment pipeline

Built a Machine Learning pipeline in Microsoft Azure Machine Learning Studio to train, evaluate, and deploy a binary classification model for predicting an individual's income in the US. The estimator used in this project is a Two-Class Boosted Decision Tree classifier. The pipeline includes

1. Data Cleaning: Substitute missing values and exclude irrelevant columns
2. Upsampling training data to account for Class Imbalance
3. Training the model and hyperparameter tuning
4. Scoring and Evaluating the Model
5. Deploying the Trained Model as a Web Service for inference

A pipeline is also created to compare the performance between one model trained on the upsampled data and the other with the original pre-processed data.

Deep Neural Network for Image Classification

Built a two-layer neural network and an L-layer neural network from scratch. Implemented all the building blocks of a neural network for image classification model:

1. Forward propagation
2. Backward propagation
3. Logistic loss
4. Parameters update using gradient descent

MySQL-Excel Application

Implementation of MySQL-Excel connection using VBA. Buttons are created to Insert, Update, and Delete MySQL data in Excel.

Recommendation Systems for movie

Developed 4 different types of recommendation systems using data from The Movie Database (TMDb) to provide relevant movie suggestions through unique filtering processes.

1. Simple Recommender
2. Content Based Recommender
3. Collaborative Filtering
4. Hybrid Engine

Data Mining Algorithms and Techniques

Plot of Linear Regression Model

Decisoon Tree model of train data for Decision tree classification

Implementation and application of Data Mining algorithms and techniques in R.

1. Decision trees: decision tree induction, overfitting, evaluating performance, comparing classifiers
2. Classification: Linear Regression, rule-based classifiers, nearest-neighbor classifiers, support vector machines, ensemble methods
3. Association analysis: frequent itemset generation, rule generation, market-basket analysis, compact representation, evaluation, categorical attributes

Cinema Management System

Created a complete cinema management system developed using Java, JavaFx and MySQL. Consists 4 sub-menus (Session, Movie, Seating, Ticketing).

Web-based Chat App

Custom chat room built with React JS (Frontend) and Firebase (Backend). Integrated Google Sign-In for secure authentication.

PDF to Audiobook App

Created an offline desktop app that transforms your PDF file into natural-sounding audio, with the option of downloading the audio in MP3 format.

Random Recipe Generator Website

Created a web application that generates a random recipe from one of the 32 countries at the click of a button.