GuyHacohen

PostDoc @ KU Leuven.
My research is focused on learning dynamics and data-efficient training, including active, continual and curriculum learning.

[Zoe Piran @zoe_piran]

Interested in revealing hidden biological signals in your single-cell data? 🪄
Make sure you add our manuscript (w/ @mor_nitzan) on SiFT, Signal Filtering for single-cell data, to your reading list📚, and also check > https://t.co/9fAnbmWxif

[Avihu Dekel @AvihuDekel]

Most active learning methods work well in high budgets, and fail in low budgets. But.. why? (Accepted to #ICML2022 🎉)
📢Active learning on a Budget – Opposite Strategies Suit High and Low Budgets
Paper: https://t.co/IOCSo9BvPd
Joint work with @GHacohen and Daphna Weinshall

[Leshem (Legend) Choshen 🤖🤗 @ACL @ICML @LChoshen]

Recycling Finetuned models, it works!

Finetuned models lie everywhere,
there must be a way to use the data and compute invested in them.

Apparently averaging their weights is such a method.

3 papers & A🧵

https://t.co/c8YIYzssT0

[Leshem (Legend) Choshen 🤖🤗 @ACL @ICML @LChoshen]

About generalization of different networks

Main finding: Generalization in pretraining follows a single dimension
Different networks, architectures, seeds, sizes but:
Similar performance → similar linguistic capabilities

@aclmeeting accepted (#NLProc)

Summary & story 🧵

[GuyHacohen @GHacohen]

כמעט עובר את מבחן טיורינג! (הצלחתי 8/12)

[Moran Mizrahi @moranmiz]

לכבוד יום השפה העברית קבלו בתרועות את אֱלִיעֶזֶר בּוֹט יְהוּדָה, האלגוריתם שמוצא חלופות עבריות למילים לועזיות באופן אוטומטי! 🎉🎊

עבור המונחים הבאים - התדעו לזהות אילו מהמילים היא החלופה העברית שקבעה האקדמיה ללשון ואיזו מהווה תוצר ממעבד קודח של מחשב? 😉
@StavYardeni @HyadataLab

[Moran Mizrahi @moranmiz]

Excited to present my latest work with @HyadataLab at @CIKM2021!
8:40pm today and 8:40am tomorrow EDT.
Full text: https://t.co/eLmMyaaTR7

We suggest a general approach to summarize procedural texts (e.g., recipes, manuals, how-to instructions, etc.) into an intuitive graph.

[GuyHacohen @GHacohen]

@moranmiz @StavYardeni אני מסרב לקבל שאני נהג שודים. אולי את סבתא

[Aviv Slobodkin @NeurIPS @lovodkin93]

Semantics in Machine Translation?
We present the first semantically-aware transformer-based NMT model.
#ML #NMT #NLP #semantic #Transformers
@LChoshen @AbendOmri
Paper: https://t.co/qP28esE2a5
In addition, here’s a short video summarizing our paper:
https://t.co/rlJeCwT5nS

[Elad Granot @eladgranot]

You were very patient, so without further ado, I'm happy to announce the new semester of the #MachineLearningClub, starting with a great talk by @GHacohen. Join us this Thu 10:30 IST, more info: https://t.co/7ZioRbUSGS

#ML @CseHuji

[Leshem (Legend) Choshen 🤖🤗 @ACL @ICML @LChoshen]

LMs learn generalizations in the same order (which?).

Same acquisition of grammatical phenomena capabilities,
regardless of data, seed, architecture.

Humbly, fascinating work with @saksheli Weinhsall @AbendOmri
https://t.co/uBxI3BlvNl
#NLProc #MachineLearning #deepRead
1/n

[GuyHacohen @GHacohen]

Another zoom session starts in 15 minutes!

https://t.co/i0vvzxJlnn

[GuyHacohen @GHacohen]

Published new paper at ICML 2020!

Let's Agree to Agree: Neural Networks Share Classification Order on Real Datasets

https://t.co/tXxWYTnvgN

See video:
https://t.co/ecnMs4Tpko

or come to the zoom session starting in 30 min:

https://t.co/i0vvzxJlnn