Forthcoming events

This page lists C2D3 events, University events, as well as related external conferences and events of interest to our members.

C2D3 Computational Biology Annual Symposium 2024
C2D3 event
Wednesday, 15 May 2024, 9.45am to 5.00pm

We warmly invite you to the C2D3 Computational Biology Annual Symposium 2024!

This event is open to everyone in the Computational Biology Community.

Packaging and Publishing Python Code for Research workshop
University of Cambridge event
Wednesday, 1 May 2024, 9.00am to 5.00pm

Would you like to learn how to package and share your code? The Accelerate Programme are planning a one day workshop to equip researchers with knowledge of workflows and tools they can use to package and publish their code. Participants will have the opportunity for hands on experience packaging and publishing a project.

University of Cambridge event
Monday, 13 May 2024, 9.30am to Wednesday, 15 May 2024, 5.00pm

This award winning course is intended to provide a strong foundation in practical statistics and data analysis using the R software environment. The underlying philosophy of the course is to treat statistics as a practical skill rather than as a theoretical subject and as such the course focuses on methods for addressing real-life issues in the biological sciences.

There are three core goals for this course:

7th Cambridge International Conference on Machine Learning and AI in (Bio)Chemical Engineering
University of Cambridge event
Tuesday, 2 July 2024, 10.00am to Wednesday, 3 July 2024, 5.00pm

02-03 July 2024
Main conference In person-only event

Paleo workshop
C2D3 event
Monday, 8 July 2024, 9.00am to Friday, 12 July 2024, 5.00pm

Co-organisers: Dr. J. Andrés Christen (CIMAT), Dr. Maarten Blaauw (Queen's University Belfast), Dr. Joan-Albert Sánchez-Cabeza (UNAM), Dr. Ana Carolina Ruiz Fernández (UNAM) and Dr. Lysanna Anderson (USGS)

Welcome to the PaleoStats Workshop: AI and Statistical Innovations for Palaeoecological Research

Forthcoming talks

A collation of interesting data science talks from across the University.

Tagging Anglo-Saxon Stone Sculptures Using Multi-Label Image Classification ML Techniques

Thursday, 25 April 2024, 2.00pm to 3.00pm
Speaker: Zeynep Aki - RSE, University of Durham
Venue: West 2, West Hub

This project involves developing a machine learning model to automatically classify images of Anglo Saxon Stone Sculptures based on their features, referred to as "tags". The aim is to have a model that can accurately identify various characteristics from these sculptures, such as animals, patterns, and architectural details, in images it has not seen before.

The process begins with data preparation, where images and associated metadata are standardized to ensure uniformity and relevance. This step involves converting images to a common format, organizing them systematically, and refining the metadata to align with the model's needs. This preparation is crucial as it directly impacts the model's ability to learn and generalize from the training data.

Following data preparation, the project employs Convolutional Neural Networks (CNNs) for the training phase. CNNs are chosen for their effectiveness in image recognition tasks. The training involves adjusting the model to identify and learn from the patterns and features in the training dataset. This includes resizing images for consistency, specifying model architecture with layers designed for feature extraction and classification, and selecting optimization and loss functions appropriate for a multi-label classification task.

This project showcases the potential of applying advanced machine learning techniques to cultural heritage preservation, offering a novel tool for cataloging and studying historical artifacts. It illustrates how technology can aid in the detailed analysis of cultural artifacts, providing deeper insights and facilitating easier access to information about our historical heritage.

Large language models for enabling constructive online conversations

Thursday, 25 April 2024, 5.00pm to 6.00pm
Speaker: Kristina Gligorić, Stanford University

NLP systems promise to disrupt society through applications in high-stakes social domains. However, current evaluation and development focus on tasks that are not grounded in specific societal implications, which can lead to societal harm. There is a need to evaluate and mitigate the societal harms and, in doing so, bridge the gap between the realities of application and how models are currently developed.
In this talk, I will present recent work addressing these issues in the domain of online content moderation. In the first part, I will discuss online content moderation to enable constructive conversations about race. Content moderation practices on social media risk silencing the voices of historically marginalized groups. We find that both the most recent models and humans disproportionately flag posts in which users share personal experiences of racism. Not only does this censorship hinder the potential of social media to give voice to marginalized communities, but we also find that witnessing such censorship exacerbates feelings of isolation. We offer a path to reduce censorship through a psychologically informed reframing of moderation guidelines. These findings reveal how automated content moderation practices can help or hinder this effort in an increasingly diverse nation where online interactions are commonplace.
In the second part, I will discuss how identified biases in models can be traced to the use-mention distinction, which is the difference between the use of words to convey a speaker's intent and mention of words for quoting what someone said or pointing out properties of a word. Computationally modeling the use-mention distinction is crucial for enabling counterspeech to hate and misinformation. Counterspeech that refutes problematic content mentions harmful language but is not harmful itself. We show that even recent language models fail at distinguishing use from mention and that this failure propagates to downstream tasks. We introduce prompting mitigations that teach the use-mention distinction and show that they reduce these errors.
Finally, I will discuss the big picture and other recent efforts to address these issues in different domains beyond content moderation, including education, emotional support, and public discourse about AI. I will reflect on how, by doing so, we can minimize the harms and develop and apply NLP systems for social good.

Title to be confirmed

Friday, 26 April 2024, 1.00pm to 2.00pm
Speaker: Speaker to be confirmed
Venue: SS03, William Gates Building.

Abstract not available

Programmed evolution: Using asexual gene drives to sculpt tumor populations and combat genetic diversity

Monday, 29 April 2024, 1.30pm to 2.30pm
Speaker: Justin Pritchard, Penn State College of Engineering
Venue: CRUK CI Lecture Theatre

Tumor heterogeneity is profound, and it provides a remarkable substrate for evolution. Despite this tremendous heterogeneity, single drugs targeting single oncogenic driver mutations can create deep responses in patients. However, these responses are ultimately lost due to the evolution of drug resistance. What if a tumor’s astonishing capacity for evolution could be hijacked to our benefit? Towards this idea, we recently developed a selection gene drive system that is stably introduced into cancer cells and is composed of two genes, or switches, that couple an inducible fitness advantage to a shared fitness cost. Using stochastic models of evolutionary dynamics, we developed design criteria for effective selection gene drives. We then build prototypes that harness the selective pressure of multiple approved tyrosine kinase inhibitors by inducing drug resistance and then deploying a second “trojan horse” switch to collapse a heterogeneous population using mechanisms as diverse as prodrug catalysis and immune activity induction. Using saturation mutagenesis and genome-wide sgRNA libraries, we show that the dual-switch selection gene drives constitute a simple motif for evolutionary control that can eradicate diverse forms of genetic resistance in vitro. Finally, using models to guide treatment scheduling, we demonstrate that carefully controlled switch engagement starting in a small fraction of cells (10% or less) can eradicate tumors in vivo.

The UK AI Safety Institute

Tuesday, 30 April 2024, 2.00pm to 3.00pm
Speaker: Nitarshan Rajkumar (University of Cambridge & UK AI Safety Institute)
Venue: Lecture Theatre 2, Computer Laboratory, William Gates Building

This talk will present an overview of efforts the UK government has been taking on AI over the past year, including the AI Research Resource, the AI Safety Summit, and with a focus on the AI Safety Institute (AISI). AISI is the world’s first state-backed organization focused on advanced AI safety for the public benefit, and is working towards this by bringing together world-class experts to understand the risks of advanced AI and enable its governance.

"You can also join us on Zoom":