Are you passionate about Mathematics, Machine Learning, and their real-world applications? Start 2025 on an inspiring note by joining us for the final talk of the Autumn semester in the Math+ML+X series!
This seminar concludes our Autumn schedule with an insightful discussion that highlights the intersection of mathematics, machine learning, and practical applications. It’s the perfect opportunity to reflect on the progress we’ve made and look ahead to the exciting topics coming in Spring 2025.

Don’t miss this opportunity to start the year with inspiration and valuable insights. Join us!
See you in Spring 2025 – we’ll be back with more engaging seminars!
 
Speakers: Meirui Jiang (The Chinese University of Hong Kong)


Title: Federated Learning for Medical Image Computing – From Algorithm Trustworthiness to Applications

Abstract: With the rapid development of big data and artificial intelligence, intelligent medical image computing has advanced significantly in recent years, becoming a revolutionary tool for enhancing clinical diagnostic and treatment efficiency. However, acquiring medical big data remains a challenging task, as individual medical institutions often struggle with limited sample sizes and incomplete case coverage.
To address these limitations, current research trends are increasingly focused on multi-center collaboration, joint training, and validation of deep learning models. Yet, the growing emphasis on patient data privacy has made multi-center data sharing more complex and restrictive. Federated learning has emerged as a promising paradigm to overcome these challenges, enabling collaborative model development across institutions. In this presentation, Dr. Jiang will discuss recent works in federated learning for medical image computing, exploring key challenges in ensuring trustworthiness, such as improving model robustness, enhancing generalization capabilities, and safeguarding privacy. The presentation will also delve into the application and validation of federated learning models using real-world clinical data.

Date & Time: January 13th, 2025 @ 4pm
Platform: Voov (Tencent)：383-870-241

Poster: Math+ML+X_Seminar_Jan_13.pdf

Are you passionate about Mathematics, Machine Learning, and their real-world applications? Start 2025 on an inspiring note by joining us for the first seminar of the year in the Math+ML+X series! This exciting event kicks off a new season of seminars, featuring engaging talks that explore new topics on mathematics and machine learning. Let’s welcome the new year with fresh ideas and amazing discussions. Don’t miss it!

Speakers: Guibin Zhang (Tongjin University)

Title: Orchestrating LLM-based Multi-agent Systems with Graph Neural Networks

Explore how novel techniques in Graph Neural Networks are revolutionising the way large language models (LLMs) work together in multi-agent systems. From coordination to real-world applications, this talk is a must-attend for anyone excited about the intersection of AI, machine learning, and complex systems. Don’t miss out!

Date & Time: January 6th, 2025 @ 3pm

Platform: Voov (Tencent)：548-600-263

Poster： Math+ML+X_January6.pdf

Are you passionate about Mathematics, Machine Learning, and their real-world applications? Join us for the grand finale of the 2024 Math+ML+X seminar series with a dual seminar featuring two exciting talks! This marks the close of 2024, with even more seminars coming in 2025. Don't miss it!

Speakers:
   •    Yanqi Cheng (University of Cambridge)


Title: You KAN Do It in a Single Shot: Plug-and-Play Methods with Single-Instance Priors


Description: Yanqi Cheng will introduce the KAN-PnP framework, which integrates Kolmogorov-Arnold Networks for solving inverse problems with minimal data. This innovative approach demonstrates superior performance in single-shot learning tasks, ensuring stable and reliable optimisation with strong convergence properties.

   •    Chu-Wun Cheng (University of Cambridge) 


Title: Mamba Neural Operator: Who Wins? Transformers vs. State-Space Models for PDEs


Description: Chu-Wun Cheng will present the Mamba Neural Operator, a framework that bridges state-space models and neural operators. By enhancing long-range dependencies and dynamics, this method outperforms traditional techniques in solving PDEs efficiently and accurately.

Date & Time: December 30th, 2024 @ 3pm
Note-- This seminar will be fully Online


Platform: Voov (Tencent)
 Meeting ID: 701-731-446

Poster: Math+ML+X_Dec30th.pdf

Are you passionate about Mathematics, Machine Learning, and their real-world applications? Join us for our engaging Math+ML+X seminar series!

Speaker: Tian Wang, Chinese Academy of Science

Title: Latent Neural Operator for Solving Forward and Inverse PDE Problems

We are excited to announce an upcoming seminar featuring Tian Wang, who will present their recent work titled Latent Neural Operator for Solving Forward and Inverse PDE Problems. This research introduces an approach to solving partial differential equations (PDEs). Neural operators have gained significant attention for solving PDE problems from data without requiring explicit equations. However, existing methods face computational challenges when the number of sample points is large. In this seminar, the author will introduce the Latent Neural Operator (LNO) framework, which leverages a Physics-Cross-Attention (PhCA) mechanism to map input samples into a learned latent space. By learning and operating within this latent space, LNO significantly reduces computational complexity while achieving state-of-the-art accuracy for forward and inverse PDE tasks.The seminar will discuss both the forward and inverse PDE problems and demonstrate the model's flexibility and generalisation capabilities. This talk is particularly relevant for researchers and students in machine learning, numerical analysis, and applied physics.

December 23rd, 2024 @3pm

Online: Voov (Tencent): 805-563-162

In-Person: Room C548 in 双清综合楼

Poster:  Math+ML+X_Seminar_Dec23.pdf

Are you passionate about Mathematics, Machine Learning, and their real-world applications? Join us for our engaging Math+ML+X seminar series!

Speaker: Zhongying Deng, University of Cambridge

We’re excited to welcome Zhongying Deng from the University of Cambridge as our upcoming speaker. He will present his work on “NorMatch: Matching Normalizing Flows with Discriminative Classifiers for Semi Supervised Learning”.

Abstract: Semi-Supervised Learning (SSL) aims to learn a model using a tiny labeled set and massive amounts of unlabeled data. To better exploit the unlabeled data the latest SSL methods use pseudo-labels predicted from a single discriminative classifier. However, the generated pseudo-labels are inevitably linked to inherent confirmation bias and noise which greatly affects the model performance. In this work we introduce a new framework for SSL named NorMatch. Firstly, we introduce a new uncertainty estimation scheme based on normalizing flows, as an auxiliary classifier, to enforce highly certain pseudo-labels yielding a boost of the discriminative classifiers. Secondly, we introduce a threshold-free sample weighting strategy to exploit better both high and low confidence pseudo-labels. Furthermore, we utilize normalizing flows to model, in an unsupervised fashion, the distribution of unlabeled data. This modelling assumption can further improve the performance of generative classifiers via unlabeled data, and thus, implicitly contributing to training a better discriminative classifier. We demonstrate, through numerical and visual results, that NorMatch achieves state-of-the-art performance on several datasets.

Don’t miss this opportunity to explore the fascinating intersection of math and machine learning!

December 16nd, 2024 @4pm

Online: Voov (Tencent): 171-242-407

In-Person: Room C548 in 双清综合楼

Poster: Math_ML_X_Dec_16_PDF.pdf

Are you passionate about Mathematics, Machine Learning, and their real-world applications? Join us for our engaging Math+ML+X seminar series! 

Speaker: Yikang Li, Peking University

We’re excited to welcome Yikang Li from Peking University as our upcoming speaker. Yikang will present her current work on “Affine Equivariant Networks Based on Differential Invariants”.  

Don’t miss this opportunity to explore the fascinating intersection of math and machine learning! 

December 9nd, 2024 @3pm

Online: Voov (Tencent): 940-474-055

In-Person: Room C548in 双清综合楼

Poster: Math+ML+X_Seminar_Dec_9.pdf

Are you passionate about Mathematics, Machine Learning, and their real-world applications? Join us for our engaging Math+ML+X seminar series!  

Speaker: Chaoyan Huang, The Chinese University of Hong Kong (CUHK)

We’re excited to welcome Chaoyan Huang from the The Chinese University of Hong Kong as our upcoming speaker. Lipei Zhang will present her current work on “Edge-guided Low-light Image Enhancement with Inertial Bregman Alternating Linearised Minimisation”. 

Don’t miss this opportunity to explore the fascinating intersection of math and machine learning! 

December 2nd, 2024 @3pm

Online: Voov (Tencent): 448-290-601

In-Person: Room B626 in 双清综合楼

Poster:  Math+ML+X_Seminar_Dec2nd.pdf

Are you passionate about Mathematics, Machine Learning, and their real-world applications? Join us for our engaging Math+ML+X seminar series! 

Speaker: Yanqi Cheng, University of Cambridge

Were excited to welcome Yanqi Cheng from the University of Cambridge as our upcoming speaker.

Plug-and-Play (PnP) priors have changed inverse problem-solving by integrating pre-trained denoisers into iterative methods. However, the reliance on large datasets poses significant challenges. In this talk, we introduce Single-Shot PnP (SS-PnP) methods, enabling effective solutions with minimal data. Our approach integrates novel proximal denoisers and neural priors to preserve essential details while overcoming gradient limitations.  

Don’t miss this opportunity to explore the fascinating intersection of math and machine learning! 

November 25th,2024 @3pm

Room B626 in 双清综合楼

Poster: Math+ML+X+nov25.pdf 

Are you passionate about Mathematics, Machine Learning, and their real-world applications? Join us for our engaging Math+ML+X seminar series! 

Speaker: Lipei Zhang, University of Cambridge

We’re excited to welcome Lipei Zhang from the University of Cambridge as our upcoming speaker. Lipei Zhang will present his approach that integrates brain tumour growth models using Partial Differential Equations (PDEs) as a regularisation technique within deep learning frameworks. 

Don’t miss this opportunity to explore the fascinating intersection of math and machine learning! 

November 18th,2024 @3pm

Room B626 in 双清综合楼 

Poster: Math_ML_X_nov_18_2024.pdf

Are you passionate about Mathematics, Machine Learning, and their real-world applications? Join us for our engaging Math+ML+X seminar series! 

Speaker: Zhanhong Ye, Peking University

We’re excited to welcome Zhanhong Ye from Peking University as our upcoming speaker. He will present on the groundbreaking topic: "PDEformer: Towards a Foundation Model for Solving PDEs." Discover how foundation models, a major trend in the field, are being harnessed to tackle Partial Differential Equations (PDEs). 

Don’t miss this opportunity to explore the fascinating intersection of math and machine learning! 

November 11th,2024 @3pm

Room B626 in 双清综合楼

Poster:  Math+ML+X_Seminar_Series_Nov_11.pdf 

We are excited to launch the Math+ML+X Seminar Series, a new platform dedicated to exploring the synergy between mathematics, machine learning, and diverse real-world applications (X). This seminar series brings together researchers at all career stages to share new ideas, foster collaboration, and showcase the latest advancements. Join us to dive into topics at the intersection of theory and practice!

NOVEMBER THEME: AI4SCIENCE

Speakers:

Haixu Wu, A Roadmap to Practical Neural PDE Solver, Tsinghua University

ChunWun Cheng, Mamba Neural Operator, University of Cambridge 

November 4th,2024 @3pm

Room B626 in 双清综合楼

Poster: Math+ML+X_Seminars.pdf