Data Science, Technion - הנדסת נתונים ומידע,טכניון

What do you get when you marry models with -net models? A better algorithm for trace . The paper “Efficient process embedding for diffusion-based trace recovery”, co-authored by Maximilian Matyash, Avigdor Gal, and Arik Senderovich (published in the Science Journal and available here: https://rdcu.be/fmdLj) is an extended version of the best-paper award winner of ’2025. The paper describes how an encoding of a process model using in a diffusion model can assist in recovering a deterministic log from a stochastic one.

מה שאלו אתכם בראיון העבודה האחרון?

LLM Interview Questions.pdf

אתם ה 88.3%:

MIT study finds AI can already replace 11.7% of U.S. workforce Artificial intelligence can already replace 11.7% of the U.S. labor market, across finance, health care and professional services, according to MIT's study.

https://www.youtube.com/watch?v=dosXzTZFhvc

alignAI Interviews with Prof. Avigdor Gal Join us for a conversation with Professor Avigdor Gal, Benjamin and Florence Free Chaired Professor of Data Science at Technion. He explores his role with al...

https://www.linkedin.com/feed/update/urn:li:activity:7393592082818404352/

What happens when you take your final course a year early?🤔 You break the timeline, confuse your future self, and accidentally build something that learns when to stop working harder than it needs… | Nagham Omar What happens when you take your final course a year early?🤔 You break the timeline, confuse your future self, and accidentally build something that learns when to stop working harder than it needs to. Last semester, four of us: Nagham Omar, Maya Rozenshtein, Vsevolod Rusanov, and Evgeny Mishlyako...

Maximilian Matyash, A Master student of in the Faculty of Data & Decision Sciences at the – Israel Institute of Technology, together with Prof. Avigdor Gal (Technion) and Prof. Arik Senderovich (York university, Canada), won the best paper award for their work “DDTR: Trace ” at the International Conference on Process Mining ( ’2025), which took place in Montevideo, Uruguay (October 2025).
The paper uses a diffusion framework, typically supporting image reconstruction, to solve a problem of trace recovery, transforming a log of activities, where the exact activity is known only in stochastic terms, into a clear deterministic log that accurately report on real-world eventualities.

Check out the full paper here:
https://arxiv.org/abs/2510.22553

https://icpmconference.org/2025/icpm-2025-hackathon/

ICPM 2025 Hackathon Overview DateStep23 September - 10 October (AoE)Team registration19 October (AoE)Deadline pitch slide deck submission22 OctoberPitch your solution (at ICPM Industry day)At ICPM closing ceremony Discover whether your team won! What is the ICPM 2025 Hackathon? In October 2025, over the course of four....

models in support of : the paper “DDTR: Diffusion Denoising Trace Recovery”, co-authored by Matyash, Avigdor Gal and Arik Senderovich, to be presented at in Montevideo, introduces a method for recovering process logs from stochastically-known logs. DDTR is an innovative algorithm that utilizes smart encoding of log and model information to train a diffusion network in identifying the noise in a trace, allowing reconstruction that performs better than state-of-the-art.