Our Projects
Client: New York Hedgefund with $14 billion assets
NLP For Stock Prediction
Project: We built a deep learning model to extract stock/ticker associated sentiment and predictive signals from unstructured text sources (e.g. news, broker research reports and social media). This sentiment was found to strongly correlate with future stock price behaviour.
Client: Swiss Hospitals
Brain Tumor Lesion Detection And Uncertainty Estimation
Project: We designed a computer vision deep learning model to automatically segment and identify white matter lesions in 3D brain scans. More importantly, we indicated the model confidence. This allows the automated deep learning system to be used efficiently in a hybrid manner with doctors – only when the system indicates a low confidence a radiologist’s expertise is required. We published 2 papers describing our system at the International Symposium on Biomedical Imaging: here and here.
Conference: Awarded Best System at TREC Conference 2022
Conversational Assistant With Information RetrievalNLP For Stock Price Prediction
Project: Automated conversational assistants are increasingly powerful and act as a useful tool in a range of domains (medical, education and finance). However, these systems often struggle with accurate information generation. Hence, we built a system that integrates a powerful information retrieval system with powerful large language model technology to build a conversational assistant to provide precise and correct information from a corpus of domain-specific content. Refer to our published paper here.
Client: Silicon Valley Start-up, >$4m annual revenue
NLPDesigning A Chatbot With High User Retention For Stock Price Prediction
Project: The start-up’s platform offers a conversational chatbot for entertainment purposes. Rewarding the chatbot explicitly for highly engaging responses has resulted in a dramatic increase in user retention. There are currently more than 350,000 daily users. See our paper here.
