ML+X Forum: Clustering & Large Language Models in Action

Join the ML+X community on Tuesday, Nov. 7, 12-1pm, to hear from two speakers who are embarking on intriguing journeys in their respective fields. Rohan Sonthalia will dive into the world of genomics as he discusses an application of deep learning for clustering mycovirus genomic sequences. Following that, Michael Roytman will unveil a unique approach to spurring self-improvement and motivation through conversational user interfaces and reinforcement learning. Both speakers are eager to receive your feedback and ideas as they venture into uncharted territories. See the speaker lineup below for details, and please register (lunch provided) by 5pm on Friday if you plan to attend in-person!

1. Clustering of genomic sequences of mycoviruses using deep learning, Rohan Sonthalia
   My project involves clustering of genomic sequences of mycoviruses using deep learning. Currently, I am looking into transformer models in order to achieve the same. One challenge that I'm facing right now is that most transformer models have been trained on human genomics, therefore the transformer may not work efficiently on genomics of viruses. As of now, what I am trying to do is just get the “input_ids”, after tokenizing the DNA sequences into numeric ids, and then apply typical ML algorithms to it, which seems to work fine. But I would like to further improve on the model to get better accuracy. In order to do so, I’ll be looking into using the transformer model along with the tokenizer to try and classify DNA sequences, or maybe even fine tune a pre-trained model on a dataset containing sequences of viruses.
2. Spurring self-improvement and intrinsic motivation using LLMs and reinforcement learning, Michael Roytman
   I have a personal training business that uses a conversational user interface, augmented with an LLM, to communicate with my clients more efficiently. However, I want to use this UI more than a communications platform. To make a dynamic NLP system work, such as this personal training chatbot, I can use a technique where labels for healthy behavior, aka fitness progress steps, are constructed by an LLM instead of me as a coach, which solves a critical performance bottleneck. The rewards would serve as controlled forms of motivation associated with motives or goals, such as improving the appearance or receiving a tangible reward. This extrinsic motivation would create a positive feedback loop and an autonomous drive to help the individual sustain internal motivation.

Finding the Orchard View room: The Orchard View room is located on the 3rd floor of Discovery Building — room 3280. To get to the third floor, take the elevator located next to the Aldo’s Cafe kitchen (see photo). If you cannot attend in-person, we invite you to stream the event via Zoom.

Join the ML+X google group: The ML+X community has a google group it uses to send reminders about its upcoming events. If you aren't already a member of the google group, you can use this link to join. Note that you have to be signed into a google account to join the group. If you have any trouble joining, please email faciltator@datascience.wisc.edu