Ishola Babatunde Isaac

Joint object detection and tracking model for real-time tracking of various objects in x-ray images. Created a model for tracking ablation catheters to get coordinates used for 3D mapping. Created a motion sensing and correction algorithm by tracking motion sensing patches to detect and correct for movement. Create a model for automatic placement of Cryoballoon by detection 2D location and 3D projection from a single x-ray image.

Generating a high-density 3D map from very few 3D points using 3D Convolutional AutoEncoder.

Sentence Similarity model trained on Quora duplicate Question Pairs dataset to identify sentences that are similar in meaning using Siamese Bidirectional LSTM model. Given two sentences, the Siamese NLP model outputs their similarity estimate. Sentence similarity model used in designing a FAQ Bot by matching user request/questions to a database of known FAQs and returning the corresponding answer.

Identifies important words in a document and suggest words similar to the identified words. Uses TF-IDF to identify the top-k most important words in a given document and finds other words similar to the identified (important) words.

Created an AI-based conversational agent with the ability to complete a user request by performing Intent Matching, Entity Extraction, Context Awareness, as well as asking follow-up questions to obtain extra information needed to complete an action. Built using DialogFlow and deployed on RCS platform. Integrated with Google Assistant for voice-based conversation.

Tube Watch is GE Healthcare’s predictive monitoring solution for CT scanners that can help predict an impending tube failure, enabling tube replacement at a more convenient time. Tube Watch integrates artificial intelligence, machine learning, and software analytics to remotely monitor and predict failure of individual components of the machine.