
Researcher at Preferred Networks (PFN, @PreferredNet).
Tokyo-to, Japan
Joined July 2012
- Tweets217
- Following361
- Followers158
- Likes4.9K
Nontawat Ch. retweeted
Today @NatureComms: @tkmtSo (@PreferredNet) and collaborators demonstrate a #neuralnetwork potential able to handle any combination of 45 elements for #materials discovery and applicable to multiple domains. #computationalchemistry #machinelearning
nature.com/articles/s41467-0…
I graduated last September and received the Dean’s award today (2022/03/24). I have been fortunate to meet with so many wonderful people here in UTokyo. I wish them the best for their future. I would not have been able to receive this award without their support.
Nontawat Ch. retweeted
🖼 A picture is worth a thousand words - especially when the words used to describe it are wrong.
In this #MadeWithTFJS episode, @helloksong shows how adversarial.js can break any neural network in browser to change a 🐕 into a 🌭 and more.
📺 → goo.gle/3sgpGJY
Nontawat Ch. retweeted
We're going to hold a talk even "Learning theory of loss functions" on Dec 15 morning (Dec 14 evening in the US). Three Ph.D. student speakers (Jessie from CU Boulder, Yutong from UMich, and I) introduce recent work. Please check it out!
c5dc59ed978213830355fc8978.d…
Nontawat Ch. retweeted
The pre-print of our #emnlp2020 paper is now on ArXiv. arxiv.org/pdf/2010.04987.pdf
So, I would like to use this thread to explain what we did. Basically, we proposed FIND -- a framework which enables humans to debug deep text classifiers.
FIND = Feature Investigation aNd Disabling
Our work on human-in-the-loop debugging deep text classifiers (joint work with my amazing supervisors @fra_toni and @lspecia) has been accepted to appear at #emnlp2020 !! Here, we made interpretability more actionable for debugging. Stay tuned for the details. 🎉
Nontawat Ch. retweeted
Great online talk event organized by @DiadochosT! I found a lot of enjoyable future work in transfer learning :)
teapaw.org/DiadochosT/status/1314…
I am very thankful to Google for this award! I also would like to express my gratitude to my lab members, colleagues, friends, and family for their support. Thank you so much :)
Google created the PhD Fellowship Program in 2009 to recognize and support outstanding graduate students pursuing exceptional research in computer science and related fields. Today, we congratulate the recipients of the 2020 Google PhD Fellowship! goo.gle/3nt9unn
Nontawat Ch. retweeted
Fresh from the press: "An Elementary Introduction to Information Geometry" mdpi.com/1099-4300/22/10/110… @Entropy_MDPI #InformationGeometry
Nontawat Ch. retweeted
Our work "Learning from Aggregate Observations" has been accepted to #NeurIPS 2020!
We study a general *weakly-supervised learning* problem where supervision signals are given to *sets* of instances, while the goal is still to predict labels of unseen individuals.
Nontawat Ch. retweeted
Our new work is out!
Time-varying Gaussian Process Bandit Optimization with Non-constant Evaluation Time. (arXiv:2003.04691v1 [stat.ML]) ift.tt/2vNFtZE
Nontawat Ch. retweeted
This work explores how causal models can be fully exploited for domain adaptation!
Intuition: humans care about causality (partially) because causal knowledge, once discovered, applies to different systems.
---- Joint work w/ Prof. Sato and Prof. Sugiyama. Comments appreciated!
Few-shot Domain Adaptation by Causal Mechanism Transfer. arxiv.org/abs/2002.03497
Nontawat Ch. retweeted
Our paper on linear-fractional metric optimization (arxiv.org/abs/1905.12511) has been accepted to AISTATS2020! (the preprint will be updated soon)
久々に主著論文が採択されました。あと地味に今日誕生日(東海岸時間で)なので、誕生日アクセプトです。よかった、、
Totally agree! Definitely one of the best talks at #NeurIPS2019. Very well-prepared and impressive.
If there is a talk you should watch from NeurIPS, it is the @EmtiyazKhan talk on Deep Learning with Bayesian principles (slideslive.com/38921489/deep…). @EmtiyazKhan and his co-authors build a beautiful bridge from Bayesian world to Deep Learning and high-order methods.
Today at #NeurIPS2019, @nichenri1995 will present our work "On the Calibration of Multiclass Classification with Rejection" by Chenri Ni, Nontawat Charoenphakdee, Junya Honda, and Masashi Sugiyama.
Tue Dec 10th, 10:45-12:45 East Exhibition Hall B+C #224
papers.nips.cc/paper/8527-on…
Please come to our poster if you are interested!
Poster: nolfwin.github.io/assets/pos…
Paper: papers.nips.cc/paper/8527-on…
Nontawat Ch. retweeted
Tutorial starts at 8:30 am (sorry not at 9am). Slides and a paper with some of the proofs are here. Slides might take a while to download (heavy!)
emtiyaz.github.io/papers/neu…
emtiyaz.github.io/papers/lea…
Excited for the tutorial tomorrow (Dec 9) at 9am at #NeurIPS2019 If you are at the conference and would to chat, please send me an email (also, if you are interested in a post-doc position in our group at Tokyo). nips.cc/Conferences/2019/Sch…




















