Rising Stars in AI Symposium 2022

The Rising Stars in AI Symposium is geared towards young researchers (including PhD students, PostDocs, young faculty) who have recently published promising work at leading venues. There will be tens of brief presentations about papers recently accepted at major AI conferences such as NeurIPS, CVPR, EMNLP, ACL, ICML, ICLR.

All speakers will be encouraged to start with an intro for non-AI experts. Normal talks: 20 min per talk: 5 for broader intro, 15 for deeper stuff. There will be plenty of time for discussions and for social activities.

The symposium will be held in KAUST with (limited) in-person attendance. To register for the event, please fill out this registration form. Due to HSE restrictions, there is a limited number of available seats for in-person attendees in the Auditorium between Building 4 and 5. We will send confirmation emails for in-person attendance.


Register Here

Agenda

Arabian Standard Time (UTC+3)

Day 1 - Sunday, March 13th, 2022

Time Speaker Presentation
09:00 - 09:20 Prof. Juergen Schmidhuber,
Dr. Tony Chan,
Prof. Lawrence Carin
Opening Remarks
09:20 – 09:40 Konstantin Burlachenko FL_PyTorch: optimization research simulator for federated learning
09:40 – 10:00 Yazeed A. Alharbi Towards interpretable and disentangled GANs
10:00 – 10:20 Juan Camilo Perez Santamaria Certifying the Robustness of Face Recognition Against Semantic Perturbations
10:20 – 11:00 Coffee Networking Break
11:00 – 11:20 Eduard Gorbunov Extragradient Method: O(1/K) Last-Iterate Convergence for Monotone Variational Inequalities and Connections With Cocoercivity
11:20 – 11:40 Guohao Li Training Graph Neural Networks with 1000 Layers
11:40 – 12:00 Deyao Zhu Motion Forecasting with Unlikelihood Training in Continuous Space
12:00 – 13:20 Lunch Break
13:20 – 13:40 Dmitry Kovalev Lower bounds and optimal algorithms for smooth and strongly convex decentralized optimization over time-varying networks
13:40 – 14:00 Mengmeng Xu Low-Fidelity Video Encoder Pre-training for Temporal Action Localization
14:00 – 14:20 Mikhail Andronov Overview of the methods for chemical reaction prediction
14:20 – 15:00 Coffee Networking Break
15:00 – 15:20 Grigory Malinovsky ProxSkip: Breaking the Communication Complexity Barrier of Local Gradient Methods
15:20 – 15:40 Merey Ramazanova Ego4D: Around the World in 3,000 Hours of Egocentric Video
15:40 – 16:00 Michael Wand Challenges in Real-world Applications of Neural Networks
16:00 – 16:20 Coffee Networking Break
16:20 – 17:00 Panel Discussion: Current trends in AI

Day 2 - Monday, March 14th, 2022

Time Speaker Presentation
09:20 – 10:00 Sebastian Stich Keynote: Algorithms for Efficient Federated and Decentralized Learning
10:00 – 10:20 Motasem Alfarra Network robustness from pixel perturbations to input transformations.
10:20 – 10:40 Alexander Tyurin Permutation Compressors for Provably Faster Distributed Nonconvex Optimization
10:40 – 11:00 Coffee Networking Break
11:00 – 11:20 Anastasia Koloskova Decentralized optimization with heterogeneous data
11:20 – 11:40 Elnur Gasanov FLIX: A simple and communication-efficient alternative to local methods in federated learning
11:40 – 12:00 Juan Carlos Alcazar Active Speakers in Context
12:00 – 13:20 Lunch Break
13:20 – 13:40 Mher Safaryan FedNL: Making Newton-Type Methods Applicable to Federated Learning
13:40 – 14:00 Francesco Faccio Parameter-based Value Functions
14:00 – 14:20 Robert Csordas Principles of Compositionality Improve Systematic Generalization of Neural Networks
14:20 – 15:00 Coffee Networking Break
15:00 – 15:20 Imanol Schlag Fast Weight Architectures for Language Modeling and Associative Inference
15:20 – 15:40 Kazuki Irie Training a Weight Matrix to Train/Modify/Adapt Itself
15:40 – 16:00 Louis Kirsch Meta learning general-purpose learning algorithms
16:00 – 16:20 Coffee Networking Break
16:20 – 17:00 Panel Discussion: Future trends in AI

Day 3 - Tuesday, March 15th, 2022

Time Speaker Presentation
09:00 – 09:40 Bernard Ghanem The AI Initiative at KAUST
09:40 – 10:00 Egor Shulgin Shifted Compression Framework for Distributed Learning
10:00 – 10:20 Guocheng Qian Make PointNet++ Faster and More Accurate
10:20 – 11:00 Coffee Networking Break
11:00 – 11:20 Rameen Abdal Extracting Semantics, Geometry, and Appearance using GANs
11:20 – 11:40 Samuel Horvath FjORD: Fair and Accurate Federated Learning under heterogeneous targets with Ordered Dropout
11:40 – 12:00 Mattia Soldan MAD: A Scalable Dataset for Language Grounding in Videos from Movie Audio Descriptions
12:00 – 13:20 Lunch Break
13:20 – 13:40 Huda A. Alamri End-to-end joint training for video question-answering
13:40 – 14:00 Zhize Li CANITA: Faster Rates for Distributed Convex Optimization with Communication Compression
14:00 – 14:20 Dylan Ashley Iterated Supervised Learning Methods for Solving Reinforcement Learning Problems
14:20 – 15:00 Coffee Networking Break
15:00 – 15:20 Anna Fruehstueck Improving Image Synthesis through specialized InsetGANs
15:20 – 15:40 Humam Alwassel Self-Supervised Learning by Cross-Modal Audio-Video Clustering
15:40 – 16:00 Vincent Herrmann Information theoretic concepts for the perception and creation of music
16:00 – 16:20 Concluding Remarks

Organizers

Scientific Committee

Silvio Giancola

Prof. Bernard Ghanem

Administrative Team

Tagreed Khalil

Liliana Rivera