🚀 Ever noticed something just feels off? 🤔 Inspired by the understanding that our brains are wired to focus on surprises, we developed Learning via Surprisability (LvS)—a novel transformation approach for timelines that excels at detecting anomalies: 🔗 arxiv.org/pdf/2503.04502 We tested LvS on: 📡 Sensor data with hidden anomalies 🌍 Global mortality trends over multiple years 🗣️ Two centuries of U.S. State of the Union Addresses The result? LvS efficiently highlights the most important outliers, helping us see the bigger picture instead of just noise. Here is an example — detecting anomalies in 30 years of world mortality reasons:

Yukie Sano talks about a fascinating topic: attention and memory. She demonstrares attention patterns during Covid, and how we tend to forget and shift attention. #netscix2025 @NetSciX_2025

Renana Keidar from HebrewU is describing the October 7th memory project #israhci2024

@erant Eran Toch is opening the Israeli HCI conference at Tel Aviv University - with 18 accepted talks and lots of posters and demos! #israhci2024

@DrCh0le organized with the women in the Recsys team a wonderful breakfast for us in #RecSys2024

The first day of #RecSys2024 was fabulous - here with @navatintarev just before Mark Riedl’s great keynote. Looking forward to the rest of the conference and the #IntRS workshop on Friday. It's good to see the Recsys gang again!

Current state

3/8 LPA findings indicated GPT-3 introduced both unique and technically detailed ideas compared to human-generated ideas. This suggests LLMs can support both divergent thinking, by broadening the range of ideas, and convergent thinking, by refining solution details.

1/8 Our research aimed to understand how LLMs, specifically GPT-3, enhance the divergent phase of group brainwriting by enriching the idea-generation process. Results showed GPT-3 offered new perspectives but also highlighted areas for improvement.

My talk today at the AI & Cyber Conference on impersonation detection and identifying ChatGPT’s texts using our method for finding unknown known, LPA youtube.com/watch?v=zgdPhmOo…

I had a great time at #NERCCS2023. Awesome people, great talks. Getting ready for my flight back to Boston on a plane not too dissimilar to the one in the video (@SeanKelty0, thanks for the reference) youtu.be/5byy_7uBvLQ

I like the way vectors and matrices are explained here: youtu.be/f5liqUk0ZTw

Proud to share that we received the Best Research Paper Award from the Israeli Association for Information Systems, for our paper Sharing emotions: determining films’ evoked emotional experience from their online reviews link.springer.com/article/10…

>> To consider viral load, we correlate the duration of meetings with the transmitted viral load and the probability of getting infected. Results obtained on real-world proximity data determine that spreading dynamics change significantly in this case! github.com/ScanLab-ossi/covi…

Our COVID-19 Research on the analysis of competing strains is published in @SciReports. In collaboration with Yuval Shahar from BGU, the work was led by our lab RA @alexzabbey. For the full paper - rdcu.be/cPonP. Here is a short thread detailing our findings: