🔥Telegram İfşa

Our work by Olga Sigalova on predicting gene expression variation and differential expression based on genomically encoded features is out! biorxiv.org/content/10.110… Great collaboration with Eileen Furlong FurlongLab Amirreza Shaeiri

Our preprint about a novel anomaly detection method based on the adversarially robust training of autoencoders is out: arxiv.org/abs/2003.05669.

We (Amirreza Shaeiri, Rozhin) have recently come up with a simple idea on how to adversarially train deep nets on large perturbations sizes: arxiv.org/abs/2003.13370.

Congratulations to the winners of the #COLT2020 best paper awards! The best paper goes to Proper Learning, Helly Number, and an Optimal SVM Bound, by Olivier Bousquet, Hanneke, Moran, and Zhivotovskiy. arxiv.org/abs/2005.11818

Conserved genomic features predictive of #GeneExpression variation across individuals indicate mechanistic link between variation & differential expression --> bit.ly/33zqrF4 Judith Zaugg Eileen Furlong EMBL #genomics #TranscriptionalRegulation

Our newest work, driven by Olga Sigalova reveals genomic features that make a gene highly responsive to any type of environmental or genetic perturbation. Keep in mind when doing DE experiments! Gene Regulation

Great article in plain English about our recent work on genomic features that are predictive of expression variation in flies and human. Recommended read for a nice summer evening :). With Olga Sigalova Eileen Furlong and contributions from Amirreza Shaeiri

New paper on arXiv: "Efficient Mean Estimation with Pure Differential Privacy via a Sum-of-Squares Exponential Mechanism," with Sam Hopkins and Mahbod Majid. Finally resolves an open problem on my mind since April 2019 (~2.5 years ago). arxiv.org/abs/2111.12981 🧵Thread ⬇️ 1/n

Check out my talk about our paper "Efficient Mean Estimation with Pure Differential Privacy via a Sum-of-Squares Exponential Mechanism" (arxiv.org/abs/2111.12981) for STOC 2022. Based on join work with Sam Hopkins and Gautam Kamath. youtu.be/GY0tCArubAg

📢 Next Wednesday (26/10) at 10am PT, Shay Moran from the Technion and Google Research will tell us about "A Characterization of Multiclass PAC Learning." More at tcsplus.wordpress.com/2022/10/20/tcs… Register (optional) at docs.google.com/forms/d/e/1FAI… for link+reminder

Hanneke, Meng, and Shaeiri give a simple resolution to a core open question in our work "Realizable Learning is All You Need". We did not know how to adapt our algorithm to the infinite multiclass setting. They give a simple variant resolving this! openreview.net/forum?id=yTfQt…