Senior Applied Scientist @Amazon AGI. Previously @MSFTResearch, @Columbia, @behaviorsignals.

Joined October 2013
8 Photos and videos
Check out Natural Instructions v2, a benchmark with 1600 #NLProc tasks and their natural language instructions. Excited to be part of this effort!
Is it possible to solve NLP tasks by simply following instructions that define the tasks? How can we measure the progress? Excited to announce Natural Instructions v2, a collection of 1600 diverse language tasks and their expert-written instructions! 📜arxiv.org/abs/2204.07705
12
Our WALNUT benchmark for semi-weakly supervised learning is coming out very soon! #NAACL2022 #NLProc
8 Apr 2022
Our paper entitled "WALNUT: A Benchmark on Semi-weakly Supervised Learning for Natural Language Understanding" got accepted in #NAACL2022. Joint with @zzzzgq @gkaraml and @AhmedHAwadallah, from Microsoft Research and Columbia. Stay tuned!
2
16
Giannis Karamanolakis retweeted
Are you #PhDone (or close)? Would you like to live in the Washington, DC area and be part of the *exciting* @GMUCompSci dept, working on Machine Translation and #nlproc for low-resource languages? I'm looking for a postdoc (ideally starting in 2022) -- do reach out if interested!
2
52
117
I recently did a Q&A with @ColumbiaCompSci talking about my research experience and projects in #ML and #NLP: cs.columbia.edu/2021/voices-… Extra: you will also find a photo advertisement of my hometown in Greece :-)
2
18
Giannis Karamanolakis retweeted
ASTRA: Self-training with Weak Supervision by @gkaraml,@subho_mpi et al. Combines rules, unlabeled data and limited labeled with self-supervision. Great results for text classification. 👩‍💻 github.com/microsoft/ASTRA 📝aclanthology.org/2021.naacl-… #python #nlproc #naacl21 #DataScience
13
15
Excited to share our #NAACL2021 paper (w. @subho_mpi, Guoqing Zheng, and @AhmedHAwadallah at Microsoft Research) on "Self-training with Weak Supervision". paper: aclweb.org/anthology/2021.na… code: github.com/microsoft/ASTRA

1
2
16
Our weak supervision method, ASTRA, leverages domain-specific rules, unlabeled data, and few labeled data through self-training. If you are attending #NAACL2021 today, join our paper's presentation and Q&A at Session 3B. Happy to chat :)
2
Giannis Karamanolakis retweeted
highlight of my @EarthInstitute post doc has been working with @Columbia students- celebrating meeting each other IRL for the first time yesterday after a hard virtual year! Congrats to Tejit/ Johanna/Dorothee on graduating @ColumbiaCompSci thanks for helping me start my lab!
1
1
25
Giannis Karamanolakis retweeted
Excited about our new paper on estimating the causal effects of linguistic properties! E.g., does writing an email politely cause faster responses? To be presented at #NAACL2021 (arxiv.org/pdf/2010.12919.pdf) W/ @rpryzant @dallascard @victorveitch @jurafsky 🧵(1/5)

4
18
80
Giannis Karamanolakis retweeted
A Survey on Recent Approaches for Natural Language Processing in Low-Resource Scenarios This survey is a great starting point for learning about low-resource NLP, common methods, and open challenges. Work by @jannikstroetgen @MicHedderich @dklakow arxiv.org/abs/2010.12309
4
112
412
Happy to share our #emnlp2020 Findings paper (with @djhsu and Luis Gravano) on “Cross-lingual text classification with minimal resources by transferring a sparse teacher”: aclweb.org/anthology/2020.fi…. tldr; we transfer weak supervision across 18 languages with limited resources.

1
1
14
If you are attending #emnlp2020 today, join our paper's presentation and QA sessions (virtual.2020.emnlp.org/paper…) as part of the Deep Learning Inside Out (DeeLIO) workshop. Extra resources: * code: github.com/gkaramanolakis/cl… * slides: tinyurl.com/y2u23ko7
1
Giannis Karamanolakis retweeted
initial attempts: very impressive QA results (check out the coref in the gates questions!) but also has some glitches.
2
14
103
Giannis Karamanolakis retweeted
(1/2) For the first time in history (since its original showing in 472 BC) a drama (The Persians by Aeschylus) will be live-streamed from the magnificent Ancient Theater of Epidaurus. thenationalherald.com/cultur…

2
27
65
Today, I will be hosting two Q/A sessions at #acl2020nlp on TXtract (aclweb.org/anthology/2020.ac…), a neural net that scales extraction to Amazon's taxonomy with ~10k product categories. TXtract is now part of Amazon's product knowledge graph!(see #kdd2020: arxiv.org/pdf/2006.13473.pdf)

1
5
<< While completing my income tax return forms >> Oh, I'm an alien, I'm a legal alien I'm a nonresident alien in New York 🎶
2
6
Giannis Karamanolakis retweeted
19 May 2020
New blog post on reviewing policies! 2020.emnlp.org/blog/2020-05-… Special focus on spurious reviews: this time no paper should be rejected primarily for not beating SOTA, for non-English work, for being a resource paper, etc. /1
9
250
753