Events

Mar 19 - Mar 25, 2023

Making federated learning practical and confidential

Dr. Ruichuan Chen, Distinguished Member of Technical Staff and a Tech Lead, Nokia Bell Labs

Mar 23, 15:30 - 16:30

B4 L5 R5209

Federated learning

Federated learning (FL) is increasingly deployed among multiple clients to train a shared model over decentralized data. To address the privacy concerns, FL systems need to protect the clients' data from being revealed during training, and also control data leakage through trained models when exposed to untrusted domains. However, existing FL systems (with distributed differential privacy) work impractically in the presence of client dropout, resulting in either poor privacy guarantees or degraded training accuracy. In addition, existing FL systems focus on safeguarding the privacy of training data, but not on protecting the confidentiality of the models being trained, which are increasingly of high business value. In this talk, I will present two pieces of our recent work that aim to address these aforementioned issues.
Interpretation, Verification and Privacy Techniques for Improving the Trustworthiness of Neural Networks

Arnaud Dethise, Ph.D. Student, Computer Science

Mar 22, 12:30 - 14:30

B1 L3 R3119

This presentation addresses the challenges associated with trusting Neural Networks due to their black-box nature and limited ability to answer important questions on how they behave. The thesis proposes techniques that increase the trustworthiness of Neural Network models by employing approaches to overcome their black-box nature. The techniques include efficient extraction and verification of weights and decisions to ensure correctness with regards to pre-existing properties, continuous and exact explanations of the model behavior, and scalable training techniques providing strong, theoretically provable guarantees of privacy. We provide strong, approximation-free guarantees about Neural Networks, improving their trustworthiness to make it more likely that users will be willing to deploy them in the real world.
Tackling credential abuse together

Michael Reiter, James B. Duke Distinguished Professor, Departments of Computer Science and Electrical & Computer Engineering, Duke University

Mar 20, 12:00 - 13:00

B9 L2 H2 H2.

data security

Despite long-ago predictions (e.g., see Bill Gates, 2004) that other user-authentication technologies would replace passwords, passwords remain not only pervasive but have flourished as the dominant form of account protection, especially at websites such as retailers that require a low-friction user experience. This talk will describe our research on methods to tackle three key ingredients of account takeovers for password-protected accounts today: (i) site database breaches, which is the largest source of stolen passwords for internet sites; (ii) the tendency of users to reuse the same or similar passwords across sites; and (iii) credential stuffing, in which attackers submit breached credentials for one site in login attempts for the same users' accounts at another.

Mar 12 - Mar 18, 2023

Mar 5 - Mar 11, 2023

Feb 26 - Mar 4, 2023

Feb 19 - Feb 25, 2023

Visualization Methods for Graphs&Networks

Prof.Oliver Deussen, Visual Computing, University of Konstanz

Feb 20, 12:00 - 13:00

B9 L2 H2 H2

visualization force-directed layouts stress-based methods vector-based constraints

Inevitably, the projection of most graph structures on two-dimensional screens will create errors and therefore visually wrong impressions. In the past, two types of methods have been developed to minimize projection errors and distribute them in a visually pleasing way. The first group of methods, force-directed layouts, interpret the links of a graph as physical springs, while stress-based methods minimize an energy function, which aims to map graph distances faithfully.

Feb 12 - Feb 18, 2023

Feb 5 - Feb 11, 2023

Jan 29 - Feb 4, 2023

Jan 22 - Jan 28, 2023

Dec 4 - Dec 10, 2022

Reynolds-robust preconditioners for the stationary incompressible Navier--Stokes equations

Prof.Patrick Farrell, University of Oxford

Dec 5, 12:00 - 13:00

B9 L2 R2322 H1

Navier-Stokes equation Robustness

Building on the work of Schöberl, Olshanskii, and Benzi, in this talk we present the first preconditioner for the Newton linearization of the stationary Navier--Stokes equations in three dimensions that achieve both optimal complexity in of count and Reynolds-robustness. The exact details of the preconditioner varies with discretization, but the general theme is to combine augmented Lagrangian stabilisation, a custom multigrid prolongation operator involving local solves on coarse cells, and an additive patchwise relaxation on each level that captures the kernel of the divergence operator.

Nov 27 - Dec 3, 2022

Bridging the Gap between the Academia and Industry by Extending High-Quality Research to Practical Business Needs

Dr.Syed Adnan Yusuf

Nov 28, 12:00 - 13:00

B9 L2 R2322 H1

Computer Vision robotics automation IoT nlp artificial intelligence

This seminar focuses on providing the audience with the context and scope of our internship program. The program is for the young and talented graduate students with an active interest in solving real-world problems. Some of the projects that will be presented in the seminar are actively developed in Elm and include domains such as computer vision, robotics and automation, healthcare, IoT, video analytics, and NLP. The seminar will serve as a launch pad to allow students to discuss their future interests and aspirations with the speaker. It will also enable them to develop a better awareness of domains more relevant to their future research aspirations.

Nov 20 - Nov 26, 2022

On training algorithms for neural networks

Jinchao Xu, Professor, Applied Mathematics and Computational Science

Nov 21, 12:00 - 13:00

B9 L2 R2322 H1

Deep learning deep neural networks

In this talk, I will first give a convergence analysis of gradient descent (GD) method for training neural networks by relating them with finite element method. I will then present some acceleration techniques for GD method and also give some alternative training algorithms

Nov 13 - Nov 19, 2022

Parameter-Free Online Optimization: Past, Present, and Future

Francesco Orabona, Associate Professor of Electrical and Computer Engineering, Boston University

Nov 14, 12:00 - 13:00

B9 L2 R2322

Parameter-free online optimization machine learning

Parameter-free online optimization is a class of algorithms that does not require tuning hyperparameters, yet they achieve the theoretical optimal performance. Moreover, they often achieve state-of-the-art performance too. An example would be gradient descent algorithms completely without learning rates. In this talk, I review my past and present contributions to this field. Building upon a fundamental idea connecting optimization, gambling, and information theory, I discuss selected applications of parameter-free algorithms to machine learning and statistics. Finally, we conclude with an overview of the future directions of this field.

Nov 6 - Nov 12, 2022

Oct 30 - Nov 5, 2022

Everything PhD students always wanted to know about research* (*but were afraid to ask)

Oct 31, 12:00 - 13:00

B9 L2 R2322 H1

scientific research PhD student

From my experience, I will try to answer doubts and dilemmas PhD students are often faced with, in their path towards a degree. Namely, I'll discuss how advisors, colleagues, peers, reviewers and so forth, fit in the universe of a PhD student, and I will end sharing my own definition of 'excellence', as an objective to pursue.

Oct 23 - Oct 29, 2022

Applied AI for sustainable development

Prof.Evgeny Burnaev, Applied AI Center, Skolkovo Institute of Science and Technology

Oct 24, 12:00 - 13:00

B9 L2 R2322 H1

Skoltech machine learning Bayesian modeling artificial intelligence

Skoltech Applied AI center’s mission is to create AI models and frameworks for solving the problems of sustainable development of industry and economy. In my presentation, I will overview the current center's activities, applied and fundamental problem statements, and corresponding recent results.

Oct 16 - Oct 22, 2022

Inference for Longitudinal Data After Adaptive Sampling

Prof. Susan Murphy, Statistics and Computer Science and Radcliffe Alumnae Professor at the Radcliffe Institute, Harvard University

Oct 20, 15:00 - 16:00

B9 L2 H2

Reinforcement Learning

In this work, we proved statistical inference for the common Z-estimator based on adaptively sampled data. Adaptive sampling methods, such as reinforcement learning (RL) and bandit algorithms, are increasingly used for the real-time personalization of interventions in digital applications like mobile health and education. As a result, there is a need to be able to use the resulting adaptively collected user data to address a variety of inferential questions, including questions about time-varying causal effects.
We used Reinforcement Learning; but did it work?

Prof. Susan Murphy, Statistics and Computer Science and Radcliffe Alumnae Professor at the Radcliffe Institute, Harvard University

Oct 19, 16:00 - 17:00

B9 L2 H2

Reinforcement Learning Digital Behavioral Health

Reinforcement Learning provides an attractive suite of online learning methods for personalizing interventions in Digital Behavioral Health. However, after a reinforcement learning algorithm has been run in a clinical study, how do we assess whether personalization occurred? We might find users for whom it appears that the algorithm has indeed learned in which contexts the user is more responsive to a particular intervention. But could this have happened completely by chance? We discuss some first approaches to addressing these questions.

Oct 9 - Oct 15, 2022

EF-BV: distributed optimization with compressed communication

Laurent Condat, Senior Research Scientist, Computer Science

Oct 10, 12:00 - 13:00

B9 L2 R2322 H1

big data distributed computing Federated learning

In the big data era, it is necessary to rely on distributed computing. For distributed optimization and learning tasks, in particular in the modern paradigm of federated learning, specific challenges arise, such as decentralized data storage. Communication between the parallel machines and the orchestrating distant server is necessary but slow. To address this main bottleneck, a natural strategy is to compress the communicated vectors. I will present EF-BV, a new algorithm which converges linearly to an exact solution, with a large class of deterministic or random, biased or unbiased compressors.

Oct 2 - Oct 8, 2022

Sep 25 - Oct 1, 2022

Designing High-Performance and Scalable Middleware for HPC, AI, and Data Science

Prof. Dhabaleswar K. (DK) Panda, Professor, Computer Science and Engineering, The Ohio State University

Sep 26, 12:00 - 13:00

B9 L2 R2322 H1

data science Deep learning machine learning artificial intelligence

This talk will focus on challenges and opportunities in designing middleware for HPC, AI (Deep/Machine Learning), and Data Science. We will start with the challenges in designing runtime environments for MPI+X programming models by considering support for multi-core systems, high-performance networks (InfiniBand and RoCE), GPUs, and emerging BlueField-2 DPUs. Features and sample performance numbers of using the MVAPICH2 libraries will be presented. For the Deep/Machine Learning domain, we will focus on MPI-driven solutions to extract performance and scalability for popular Deep Learning frameworks (TensorFlow and PyTorch), large out-of-core models, and Bluefield-2 DPUs.

Sep 18 - Sep 24, 2022

Towards Detailed Understanding of the Visual World

Fahad Khan, Associate Professor at MBZUAI and Linköping University

Sep 19, 12:00 - 13:00

B9 L2 R2322

Computer Vision

Machine perception that corresponds to the ability to understand the visual world based on the input from sensors, such as cameras is one of the central problems in Artificial Intelligence. To this end, recent years have witnessed tremendous progress in various instance-level recognition tasks having real-world applications in e.g., robotics, autonomous driving and surveillance. In this talk, I will first present our recent results towards understanding state-of-the-art deep learning-based visual recognition networks in terms of their robustness and generalizability. Next, I will present our results on learning visual recognition models with limited human supervision. Finally, I will discuss moving one step further from instance-level recognition to understand visual relationships between object pairs.

Sep 11 - Sep 17, 2022

Sep 4 - Sep 10, 2022

Communication-Efficient Distributed Newton-type Algorithms for Federated Learning

Mher Safaryan, Postdoctoral Research Fellow, Applied Mathematics and Computational Science

Sep 5, 12:00 - 13:00

B9 L2 R2322 H1

machine learning

In this talk, I will discuss communication compression and aggregation mechanisms for curvature information in order to reduce these costs while preserving theoretically superior local convergence guarantees.