http://talks.cam.ac.uk/show/rss/5408

A survey of methods of interpretability of neural networks: from gender bias mitigation to interpreting BERT embeddings in a psycholinguistic manner.

Bio:

Giuseppe Attanasio is a postdoctoral researcher affiliated with the Milan Natural Language Processing (MilaNLP) Lab at Bocconi University. His research primarily focuses on large-scale neural architectures for Natural Language Processing.

Attanasio has contributed to various research projects and publications in the field of NLP . Notably, he has worked on topics such as automatic misogyny identification, benchmarking post-hoc interpretability approaches for transformer-based models, and entropy-based attention regularization for bias mitigation. His work often involves the development and deployment of NLP algorithms to address real-world problems .

• Speaker: Giuseppe Attanasion, Milan NLP
• Friday 17 May 2024, 12:00-13:00
• Venue: https://cl-cam-ac-uk.zoom.us/j/4361570789?pwd=Nkl2T3ZLaTZwRm05bzRTOUUxY3Q4QT09&from=addon .
• Series: NLIP Seminar Series; organiser: Richard Diehl Martinez.

Add to your calendar or Include in your list

Abstract not available

• Speaker: Carina Kauf
• Friday 07 June 2024, 12:00-13:00
• Venue: SS03, William Gates Building. Zoom link: https://cl-cam-ac-uk.zoom.us/j/4361570789?pwd=Nkl2T3ZLaTZwRm05bzRTOUUxY3Q4QT09&from=addon .
• Series: NLIP Seminar Series; organiser: Richard Diehl Martinez.

Add to your calendar or Include in your list

Abstract not available

• Speaker: Sabri Eyuboglu
• Friday 14 June 2024, 12:00-13:00
• Venue: FW26, William Gates Building. Zoom link: https://cl-cam-ac-uk.zoom.us/j/4361570789?pwd=Nkl2T3ZLaTZwRm05bzRTOUUxY3Q4QT09&from=addon .
• Series: NLIP Seminar Series; organiser: Richard Diehl Martinez.

Add to your calendar or Include in your list

Abstract not available

• Speaker: Tim Rocktäschel
• Friday 24 May 2024, 12:00-13:00
• Venue: FW26, William Gates Building. Zoom link: https://cl-cam-ac-uk.zoom.us/j/4361570789?pwd=Nkl2T3ZLaTZwRm05bzRTOUUxY3Q4QT09&from=addon .
• Series: NLIP Seminar Series; organiser: Richard Diehl Martinez.

Add to your calendar or Include in your list

Abstract not available

• Speaker: Tim Rocktäschel
• Friday 24 May 2024, 12:00-13:00
• Venue: FW26, William Gates Building. Zoom link: https://cl-cam-ac-uk.zoom.us/j/4361570789?pwd=Nkl2T3ZLaTZwRm05bzRTOUUxY3Q4QT09&from=addon .
• Series: NLIP Seminar Series; organiser: Richard Diehl Martinez.

Add to your calendar or Include in your list

A survey of methods of interpretability of neural networks: from gender bias mitigation to interpreting BERT embeddings in a psycholinguistic manner.

• Speaker: Giuseppe Attanasion, Milan NLP
• Friday 17 May 2024, 12:00-13:00
• Venue: https://cl-cam-ac-uk.zoom.us/j/4361570789?pwd=Nkl2T3ZLaTZwRm05bzRTOUUxY3Q4QT09&from=addon .
• Series: NLIP Seminar Series; organiser: Richard Diehl Martinez.

Add to your calendar or Include in your list

Among the most difficult issues in CFD is the very wide range of scales involved in some problems. Attempts at investigating the dynamics contact line have been made coming from various theoretical and numerical frameworks, the closest to first principles being molecular dynamics, while diffuse interface methods and sharp interface methods with several variants have also been put forward. Experiments are obviously difficult. Efforts made on a number of typical cases, including plunging and withdrawing plates, a sheared droplet, sessile droplets on oscillating or accelerating substrates, menisci in nanopores and the hydrodynamics assist problem. The issues involved in nucleate boiling and accelerated sessile droplets will be addressed both from the point of view of experiments (performed by various colleagues from MIT and Tokyo University) and from the point of view of simulations.

I will also show recent developments in the Basilisk code allowing to simulate contact lines on complex curved boundaries, using the immersed boundary method and an appropriate contact angle boundary condition, and inside porous media.

• Speaker: Stephane Zaleski, Institut Jean Le Rond ∂’Alembert, Sorbonne Université
• Thursday 23 May 2024, 11:30-12:30
• Venue: Open Plan Area, Institute for Energy and Environmental Flows, Madingley Rise CB3 0EZ.
• Series: Institute for Energy and Environmental Flows (IEEF); organiser: Catherine Pearson.

Add to your calendar or Include in your list

From Conspiracist to Ex-Conspiracist: Brent Lee’s Journey Out of the Rabbit Hole

• Brent Lee, a former conspiracist who was deeply entrenched in the world of conspiracy theories from 2003 to 2018, shares his personal journey of transformation and the factors that led him to question and ultimately reject his previous beliefs.

• Lee’s initial exposure to conspiracy theories began in 2003 when he stumbled upon a file-sharing application called DC++ and discovered a folder of videos related to the Illuminati and Freemasons.

• He became engrossed in the works of prominent conspiracy theorists such as Jordan Maxwell, William Cooper, David Icke, and Alex Jones, consuming their books, videos, and radio appearances obsessively.

• Lee actively engaged with other conspiracy theorists on online forums, contributing to the development of alternative narratives for events like the 7/7 London bombings and decoding hidden symbols in high-profile events.

• However, doubts began to arise when he encountered increasingly outlandish theories, such as the Sandy Hook massacre being a hoax, Justin Bieber and Eminem being Illuminati clones, and the rise of flat earth proponents.

• As his interests in politics, religion, society, and culture expanded, Lee found himself increasingly at odds with the grand conspiracy narrative and the logistics of how it could function in the real world.

• Significant global events, such as the rise of QAnon, the election of Donald Trump, Brexit, and the leadership of Jeremy Corbyn in the UK, further challenged his belief in a predetermined new world order.

https://www.tinfoilconfessions.com/my-story

• Speaker: Brent Lee
• Wednesday 22 May 2024, 15:00-16:00
• Venue: Ground Floor Lecture Theatre, Department of Psychology, Downing Site, Cambridge.
• Series: Social Psychology Seminar Series (SPSS); organiser: Yara Kyrychenko.

Add to your calendar or Include in your list

The Cancer Research UK Cambridge Centre and the Department of Radiology at Addenbrooke’s are pleased to announce a seminar series on Artificial Intelligence (AI) in Medicine, which aims to provide a comprehensive overview of the latest developments in this rapidly evolving field. As AI continues to revolutionize healthcare, we believe it is essential to explore its potential and discuss the challenges and opportunities it presents.

The seminar series will feature prominent experts in the field who will share their research and insights on a range of topics, including AI applications in disease diagnosis, drug discovery, and patient care. Each seminar will involve two talks, followed by an interactive discussion with a light lunch from Aromi! We hope that this seminar series will be a valuable platform for researchers, practitioners and students to learn about the latest trends and explore collaborations in the exciting field of AI in Medicine.

The next seminar will be held on 21 May 2024, 12-1pm at the Jeffrey Cheah Biomedical Centre (Main Lecture Theatre), University of Cambridge and streamed online via Zoom. A light lunch from Aromi will be served from 11:50. This month will feature the following two talks:

Big Data and AI in Cardiac Imaging – Making a difference? – Jonathan Weir-McCall, Assistant Professor, Department of Radiology, University of Cambridge

Jonathan is a University Lecturer at the University of Cambridge and an Honorary Consultant Cardiothoracic Radiologist at the Royal Papworth Hospital. His research interests lie in the use of cardiovascular CT and MRI for better understanding how these can be used to improve patient treatment and outcomes in structural and coronary artery disease. He has authored >130 peer reviewed publications, co-authored the SCCT guidelines on the role of CT in the assessment for transcatheter aortic valve insertion, and the BSCI /BSTI guidelines on the reporting of calcification on routine chest CT. He sits on the executive committee of the BSCI , guideline committee of the SCCT , and Diagnostic Advisory Committee of NICE .

Abstract: AI and advanced analytics are reaching clinical practice with significant opportunities, but also challenges in determining their real world impact and efficacy. In cardiac imaging, advanced analytics using AI and computational fluid dynamics are being routinely used in clinical care. While small scale randomised control trials present promising insights into their potential benefits, real world data is lacking. Leveraging national datasets we analyse the impact of these technologies in the UK, examining the impact of one AI-augmented CT tool on healthcare behaviours and patient outcomes.

Learning structures in multimodal pathology – Konstantin Hemker, PhD Candidate, Computer Laboratory, University of Cambridge

Konstantin is a PhD student in the Computer Lab at the University of Cambridge focussing on multimodal representation learning for biomedical data modalities. In particular, he is looking at how fusion models can provide multi-scale context in computational pathology. Before starting his PhD, Konstantin worked as a Senior Data Scientist in the Healthcare and Pharmaceuticals practice at the Boston Consulting Group, focussing on drug yield optimisation of active ingredients in antibody treatments and radiocontrast agents. He holds Master’s degrees in Computer Science from Imperial and Cambridge and an undergraduate degree from the London School of Economics.

Abstract: Integrative modelling of multiple data structures (such as images, graphs, sequences, or tabular data) in the same model is a common challenge for machine learning approaches in biomedical domains. This challenge arises from a lack of shared semantics between modalities, one-to-many relationships, missing modalities, and data sparsity. Meanwhile, multi-scale context can provide important information about the tumour microenvironment in fields such as computational pathology and consequently help train better predictive models. This talk will cover state-of-the-art multimodal representation learning methods that can learn from multiple data distributions, capture cross-modal relationships, and handle missing modalities whilst maintaining structural information from each modality for predictive tasks in pathology.

This is a hybrid event so you can also join via Zoom:

https://zoom.us/j/99050467573?pwd=UE5OdFdTSFdZeUtIcU1DbXpmdlNGZz09

Meeting ID: 990 5046 7573 and Passcode: 617729

We look forward to your participation! If you are interested in getting involved and presenting your work, please email Ines Machado at im549@cam.ac.uk

• Speaker: Dr. Jonathan Weir-McCall and Konstantin Hemker
• Tuesday 21 May 2024, 12:00-13:00
• Venue: Jeffrey Cheah Biomedical Centre (Main Lecture Theatre), University of Cambridge.
• Series: Cambridge MedAI Seminar Series; organiser: Ines Machado.

Add to your calendar or Include in your list

The effects of global warming are calling for a rapid transition from fossil to renewable energy sources. New materials for energy generation and storage are required to enable this transition. In this presentation we will introduce a new research tool which introduces elements of automation and artificial intelligence to the field of materials discovery. The new research platform will be hosted at the Melbourne Centre for Nanofabrication and is part of the Australian Centre for Advanced Photovoltaics. The main purpose of this installation is to accelerate the development of novel printable photovoltaic materials such as lead halide perovskites and their lead-free analogues, it is however also applicable towards the discovery of energy materials more broadly. It entails three integrated robotic sections. The first allows to formulate inks from liquid and solid chemical precursors, the second section allows to produce thin films from these inks via solution processing techniques and the third section hosts a number of characterization tools to determine the optical, electronic and structural properties of these films. The entire system operates under inert gas atmosphere with a clock speed of 5 minutes and the ability to operate autonomously for 24 hours.

• Speaker: Udo Bach - Monash University
• Tuesday 21 May 2024, 11:30-12:30
• Venue: Lecture Theatre 1, Department of Chemical Engineering and Biotechnology, West Cambridge Site.
• Series: Chemical Engineering and Biotechnology; organiser: stranks-pa.

Add to your calendar or Include in your list

Dylan O’Toole: ‘The value of neutrophil cell population data (CPD) parameters, NE-SFL and NE-SSC, as markers of systemic inflammation in dogs and cats’

• Speaker: Dylan O'Toole, Department of Veterinary Medicine
• Friday 31 May 2024, 08:45-10:00
• Venue: LT2.
• Series: Friday Morning Seminars, Dept of Veterinary Medicine; organiser: Fiona Roby.

Add to your calendar or Include in your list

Abstract not available

• Speaker: Brent Lee
• Wednesday 22 May 2024, 15:00-16:00
• Venue: Ground Floor Lecture Theatre, Department of Psychology, Downing Site, Cambridge.
• Series: Social Psychology Seminar Series (SPSS); organiser: Yara Kyrychenko.

Add to your calendar or Include in your list

Abstract not available

• Speaker: Laura M Machesky, Sir William Dunn Professor of Biochemistry, University of Cambridge
• Monday 17 June 2024, 11:00-12:00
• Venue: In person in the Max Perutz Lecture Theatre (CB2 0QH) and via Zoom, link: https://mrc-lmb-cam-ac-uk.zoom.us/j/99537530757?pwd=UVk1YjIyeFY4QzNjNkhwa0xwbERoQT09.
• Series: MRC LMB Seminar Series; organiser: Scientific Meetings Co-ordinator.

Add to your calendar or Include in your list

Human genes act together in complex networks to perform a myriad of functions. Whereas details maps exist for most biochemical pathways, the wiring of genetic networks in human cells is poorly understood. Using genetics in haploid human cells we have assigned genes to more than a hundred quantitative cellular phenotypes.This approach enables us to identify missing enzymes in cell biology, alternative metabolic routes and a pathway that activates cell death in response to cancer therapeutics.

• Speaker: Thijn Brummelkamp, Netherlands Cancer Institute/Oncode Institute
• Monday 03 June 2024, 11:00-12:00
• Venue: In person in the Max Perutz Lecture Theatre (CB2 0QH) and via Zoom, link:https://mrc-lmb-cam-ac-uk.zoom.us/j/95178919042?pwd=OUp6RUN3a0dkNFZqWGxQV0p0T0JOZz09.
• Series: MRC LMB Seminar Series; organiser: Scientific Meetings Co-ordinator.

Add to your calendar or Include in your list

Let k be a field of characteristic p>0 and let G be a topological group admitting an open pro-p subgroup. If H=k[I\G/I] denotes the corresponding Hecke algebra, then the functor of taking I-invariants provides an adjunction between the category of smooth k-linear representations of G and the category of H-modules. When H is the pro-p Iwahori-Hecke algebra of a p-adic reductive group, this adjunction is not well-understood. This is one of the main obstacles to understanding the smooth representation theory of these groups and to the mod-p Langlands program. I will describe a recent approach to studying this adjunction using the language and techniques from the theory of model categories. I will then finish by describing the homotopy classes of simple supersingular pro-p Iwahori-Hecke modules for GL_n.This is joint work with Jan Kohlhaase

• Speaker: Nicolas Dupré, University of Duisburg-Essen
• Wednesday 29 May 2024, 16:30-17:30
• Venue: MR12.
• Series: Algebra and Representation Theory Seminar; organiser: Adam Jones.

Add to your calendar or Include in your list

Learning problems augmented with latent symmetries have attracted considerable interest in recent years. A significant class of such problems arises in experiments where a system is constrained to evolve in accordance with the rigid laws of nature, such as the celebrated technique of cryo electron microscopy (Cryo-EM). The constraint of such latent symmetries, given by group invariances or equivariances, precludes the possibility of having many repeated measurements of the exact same object, and poses a fundamental challenge for learning a signal in the presence of ambient noise. We will start with a gentle introduction to the problem of learning under latent symmetries, and explore its intriguing connections with a range of disparate topics — invariant theory, harmonic analysis, compressive sensing and Gaussian calculus. We will subsequently specialise to the Multi Reference Alignment (MRA) model, and explore the fundamental aspects of the recovery problem (such as sample complexity) in the presence of structural constraints on the signal (such as sparsity). In particular, we unveil a novel quartic dependence on noise level for the sample complexity of sparse MRA , leveraging a range of mathematical tools from uncertainty principles of Fourier analysis to techniques from combinatorial optimisation.

Based in part on the following works :

[1] Sparse Multi-Reference Alignment: Phase Retrieval, Uniform Uncertainty Principles and the Beltway Problem, S. Ghosh and P. Rigollet, Foundations of Computational Mathematics, 23(5), pp.1851-1898 (2023)

[2] Minimax-optimal estimation for sparse multi-reference alignment with collision-free signals, S. Ghosh, S.S. Mukherjee, J.B. Pan, arXiv preprint

• Speaker: Subhro Ghosh, National University of Singapore
• Friday 31 May 2024, 14:00-15:00
• Venue: MR12, Centre for Mathematical Sciences.
• Series: Statistics; organiser: Dr Sergio Bacallado.

Add to your calendar or Include in your list

As theoretical physicists we know that Mathematics is unreasonably effective in describing our dynamical systems. This effectiveness singles out relevant mathematical structures in optimisation searches for models. In this talk I give an overview on where we can successfully apply this to search for symmetries or integrable structures of dynamical systems. Then I discuss how we use these methods to explore mathematical structures in beyond the Standard Model physics. In particular, we will use the string theory landscape as a benchmark for ML methods to search for new structures in this space.

• Speaker: Sven Krippendorf (LMU)
• Friday 17 May 2024, 16:00-17:00
• Venue: MR19 (Potter Room, Pavilion B), CMS.
• Series: HEP phenomenology joint Cavendish-DAMTP seminar; organiser: Benjamin Christopher Allanach.

Add to your calendar or Include in your list

Energy metabolism is essential for providing the energy necessary to ensure proper cellular function. Mutations in genes regulating this process lead to inherited metabolic disorders that can particularly affect tissues with high energy demands like the brain. The limited access to patient neural tissue and the difficulty to manipulate mitochondrial DNA complicates the development of transgenic animal models and cellular models, which are needed for treatment dis-covery. In this talk, I will summarize our efforts in using patient-derived and engineered induced pluripotent stem cells (iPSCs) to study mitochondrial neurological diseases. I will show examples of this approach in the context of Huntington´s disease (HD) and Leigh syndrome (LS), which is the most frequent and most severe mitochondrial disease affecting 1/40,000 newborns. We show that neuronal cultures and brain organoids derived from HD and LS-specific iPSCs can be used as model systems to investigate the neuropathological mechanisms and to carry out phenotypic compound screenings. Our data pave the way to the identification of disease-modifying therapies for currently incurable neurological and mitochondrial disorders.

• Speaker: Professor Dr. Alessandro Prigione
• Wednesday 05 June 2024, 15:00-16:00
• Venue: MRC MBU, Level 7 Lecture Theatre, The Keith Peters Building, CB2 0XY.
• Series: MRC Mitochondrial Biology Unit Seminars; organiser: Lisa Arnold.

Add to your calendar or Include in your list

The next event in our popular Seminar Series will take place on Wednesday, 5th June.

This seminar will include time for Q&A and will be followed by refreshments and networking.

Focus on: targeted protein degradation

Speakers: Heike Laman, Department of Pathology, University of Cambridge “Characterization of nanobodies against N-Myc for drug development in the treatment of Neuroblastoma”

and

Benedict Cross, Phoremost “Engineering degradation: finding new ways to destroy undruggable targets”

Register at: https://www.tickettailor.com/events/milnerinstitute/1243021

• Speaker: Heike Laman, University of Cambridge and Benedict Cross, Phoremost
• Wednesday 05 June 2024, 16:00-17:00
• Venue: Lecture Theatre, Jeffrey Cheah Biomedical Centre.
• Series: Milner Seminar Series; organiser: Mary-Jane Roebuck.

Add to your calendar or Include in your list

The next event in our popular Seminar Series will take place on Wednesday, 5th June.

This seminar will include time for Q&A and will be followed by refreshments and networking.

Focus on: targeted protein degradation

Speakers:

Heike Laman, Department of Pathology, University of Cambridge

“Characterization of nanobodies against N-Myc for drug development in the treatment of Neuroblastoma”

and

Benedict Cross, Phoremost

“Engineering degradation: finding new ways to destroy undruggable targets”

Register at: https://www.tickettailor.com/events/milnerinstitute/1243021

• Speaker: Heike Laman, University of Cambridge and Benedict Cross, Phoremost
• Wednesday 05 June 2024, 16:00-17:00
• Venue: Lecture Theatre, Jeffrey Cheah Biomedical Centre.
• Series: Milner Seminar Series; organiser: Mary-Jane Roebuck.

Add to your calendar or Include in your list