Integrated Calendar

Over the last decade, several exciting directions have been initiated by laser-driven plasmas,
e.g., compact particle accelerators, inertial fusion and laboratory astrophysics. The first has
known rapid progress, in terms of current, energy, stability; fusion has gone through a historic
step, with the news of ignition being achieved at NIF in 2022; and laboratory astrophysics has
known also spectacular developments, demonstrating the possibility to perform fully scalable
experiments relevant to various objects such as forming stars and supernovae. A particularly
interesting aspect is that all these fields are strongly synergistic, i.e., that advances in one can
push the others as well. I will present examples of such synergies, through recent results
we have obtained in all these domains, and in particular how ultra-bright neutron beams
can be developed using latest generation multi-PW lasers [1,2]. These could open interesting
perspectives in terms of cargo inspection, but also for fusion plasma measurements.
I will also show how fusion can benefit from external magnetization [3]. Finally, I will discuss
advances in laboratory astrophysics, particularly the first-stage acceleration of ions leading to
cosmic rays [4,5], understanding the universal nature of collimated outflows in the Universe [6],
and probing the intricacy of 3D magnetic reconnection [7]
[1] High-flux neutron generation by laser-accelerated ions from single-and double-layer targets, V Horný et al.,
Scientific Reports 12 (1), 19767, 2022
[2] Numerical investigation of spallation neutrons generated from petawatt-scale laser-driven proton beams,
B Martinez et al., Matter and Radiation at Extremes 7 (2), 024401, 2022
[3] Dynamics of nanosecond laser pulse propagation and of associated instabilities in a magnetized underdense
plasma, W. Yao et al., https://doi.org/10.48550/arXiv.2211.06036
[4] Laboratory evidence for proton energization by collisionless shock surfing, W Yao et al.,
Nature Physics 17 (10), 1177-1182, 2021
[5] Enhancement of the Nonresonant Streaming Instability by Particle Collisions, A Marret et al.,
Physical Review Letters 128 (11), 115101, 2022
[6] Laboratory disruption of scaled astrophysical outflows by a misaligned magnetic field, G Revet et al.,
Nature communications 12 (1), 762, 2021
[7] Laboratory evidence of magnetic reconnection hampered in obliquely interacting flux tubes, S Bolaños et al.,
Nature Communications 13 (1), 6426, 2022

LLMs offer users intuitive interfaces for interacting with textual information. The integration of vision into LLMs through VLMs has enabled these models to "see" and reason over visual content. However, these VLMs possess generic knowledge, lacking a personal touch. This raises an intriguing question: can we equip these models with the ability to comprehend and utilize user-specific concepts, tailored specifically to you? Can we ask the model questions about you, such as what you are wearing or what your friend is doing in the image? In this talk, we will explore how we can personalize VLMs to each user, offering more meaningful interactions that better reflect individual experiences and relationships.

Bio:

Yuval is a PhD student at Tel Aviv University under the supervision of Prof. Daniel Cohen-Or. His research centers around leveraging generative models to give users greater control and creative freedom when interacting with visual content. Currently interning at Snap Research under Kfir Aberman, Yuval is also exploring new approaches for personalizing generative models.

As machine learning models get more complex, they can outperform traditional algorithms and tackle a broader range of problems, including challenging combinatorial optimization tasks. However, this increased complexity can make understanding how the model makes its decisions difficult. Explainable models can increase trust in the model’s decisions and may even lead to improvements in the algorithm itself. Algorithms like GradCAM or SHAP provide good explanations in terms of feature importance, typically for classification tasks. Still, they provide little insight when the ML pipeline is designed to work, for example, as an algorithm for solving optimization problems.                                                                      In this talk, we present a concept-learning framework for explaining a neural machine-learning model’s decision-making process from an algorithmic point of view. Using the NeuroSAT algorithm for SAT solving as a case study, we demonstrate how our framework finds the algorithmic concepts that drive the operation of NeuroSAT. Using the concepts that we discover, we can re-write the black box NeuroSAT net as a text-book algorithm that performs typical algorithmic moves like (a) compute confidence levels for every variable, (b) fix variables with the highest confidence and simplify the instance, (c) solve the residual formula using some simple technique. (Such a principle guides, for example, the well-known Belief-Propagation-Decimation algorithm).

Joint work with Elad Shoham (PhD student BGU), Kahalil Wattad (MSc student BGU), Hadar Cohen (MSc student BGU), and Havana Rika (Tel-Aviv-Yafo Academic College). 

Short bio:

Dan Vilenchik holds a PhD in computer science from Tel Aviv University. He did a postdoc at UC Berkeley and UCLA. He is currently a tenured member of the Electrical Engineering School at Ben-Gurion University. His research includes various aspects of machine learning, such as the challenges of high-dimensional data, explainable AI, NLP, and multidisciplinary projects.

 

TBD

Vizing’s Theorem provides an algorithm that edge colors any graph of maximum degree Δ using Δ 1 colors, which is necessary for some graphs, and at most one higher than necessary for any graph. In online settings, the trivial greedy algorithm requires 2Δ-1 colors, and Bar-Noy, Motwani and Naor in the early 90s showed that this is best possible, at least in the low-degree regime. In contrast, they conjectured that for graphs of superlogarithmic-in-n maximum degree, much better can be done, and that even (1 o(1))Δ colors suffice online. This would make edge coloring a rare problem, for which "online is (nearly) as easy as offline". In this talk I will outline the history of this conjecture, and its recent resolution, together with extensions of a flavor resembling classic and recent results on *list* edge-coloring and “local” edge-coloring.

Talk based in part on joint works with many wonderful and colorful collaborators, including Sayan Bhattacharya, Joakim Blikstad, Ilan R. Cohen, Fabrizio Grandoni, Seffi Naor, Binghui Peng, Amin Saberi, Aravind Srinivasan, Ola Svensson and Radu Vintan.