Research Projects BCI

Here are a couple of projects building towards the generation of novel BCI experiments. Several other BCI projects are available in collaboration with specific labs. Please reach out to know more.

Next-Gen Brain-Machine Interfaces: Real-Time Spike Sorting with Neuropixels 

This project offers a unique opportunity to develop novel brain-machine interface (BCI) technology. The goal is to create a real-time spike sorting algorithm for Neuropixels probes, which record an unprecedented number of neurons and enable long-term (months) recordings in freely moving mice. This groundbreaking technology, already approved for human use, allows large-scale continuous neural monitoring, opening new avenues for understanding brain function. By leveraging deep learning, the project aims to manage this vast data stream, enabling novel BCI tasks where animals learn to modulate their neural activities. Success would mark a major achievement in BCI applications.

Are LLMs errors and logical vulnerabilities human-like? 

This project investigates Large Language Models (LLMs), focusing on their errors and vulnerabilities. The goal is to understand the conditions and inputs that lead these models to make mistakes and identify ways to exploit these errors. A key aspect is comparing LLMs’ mistakes with human errors. The project will categorize LLM errors and design tasks for human participants to determine if humans make similar mistakes under the same conditions. These tasks will mirror the conditions that lead LLMs to err, allowing for a direct comparison between human cognitive reasoning and LLM responses. The findings could provide valuable insights into LLM limitations and contribute to developing more robust models

Cell-type classification via predictive neural embeddings

This project tackles a critical challenge in neuroscience: predicting the genetic information of cells from their neural activity. This classification problem involves identifying cell types based on neural activity. By utilizing neural activity embeddings, we will categorize neurons according to their genotype leveraging newly recorded and available datasets. We aim to develop state-of-the-art tools that can be used in real-time BCI applications. This project lies at one of the most important scientific intersections: the crossroad between neuroscience and genomics.