iiinova — where your digital self keeps evolving

Zhaoyang Wang @zhaoyangw.bsky.social · 05/09/2026

AI Agent Product Design Proactive Agent: A proactive agent can work continuously and regularly and try to discover and solve new problems on its own. Wrong Agent Design: My Agent can do anything. Correct Agent Design: In areas A, B, and C, my Agent can do a much better job.

020

Reposted by Zhaoyang Wang

Jeffrey P. Bigham @jeffreybigham.com · 13/08/2026

model capability jaggedness is part of the unintuitiveness of current AI, but it's made even less intuitive by tirelessness… wigguming through the jaggedness toward something that looks like success. not quite a paperclip factory, but not so far off. metaphor for life? openai.com/index/huggin...

openai.com

OpenAI and Hugging Face partner to address security incident during model evaluation

OpenAI and Hugging Face share early findings from a security incident during AI model evaluation, highlighting advanced cyber capabilities and lessons for defenders.

031

Reposted by Zhaoyang Wang

Alexander Terenin @avt.im · 17/07/2025

At ICML, we're presenting a paper on uncertainty-aware surface reconstruction! Compared to previous approaches, we are able to completely remove the need for recursive linear solves for reconstruction and interpolation, using geometric GP machinery. Check it out!

1133

Reposted by Zhaoyang Wang

Yoshua Bengio @yoshuabengio.bsky.social · 10/07/2025

The Code of Practice is out. I co-wrote the Safety & Security Chapter, which is an implementation tool to help frontier AI companies comply with the EU AI Act in a lean but effective way. I am proud of the result! 1/3

1256

Reposted by Zhaoyang Wang

Jannis Bulian @j5b.bsky.social · 17/06/2025

The Gemini 2.5 Technical Report is out: storage.googleapis.com/deepmind-med...

0112

Reposted by Zhaoyang Wang

Neil Lawrence @lawrennd.bsky.social · 22/05/2025

I first spoke at the #TheAISummit in 2022 at a panel session promoting the CDEI Federated Learning Sandbox. This year I'll be on the Headliners Stage on 11th of June talking about strategic decision making in the age of AI. london.theaisummit.com/speakers/nei...

083

Reposted by Zhaoyang Wang

Michael Bronstein @mmbronstein.bsky.social · 13/05/2025

Erlangen AI Hub, unlike the name suggests, is located in England. Join us for a cool conference at the intersection of math and ML.

0102

Reposted by Zhaoyang Wang

Neil Lawrence @lawrennd.bsky.social · 12/05/2025

Tomorrow, Tuesday 13th May I'll be on Radio 4 from 9 am for Life Scientific. It will also be available online via BBC Sounds from the first broadcast: www.bbc.co.uk/programmes/b... Thank you @jimalkhalili.bsky.social and #bbcradio4 for inviting me!

1135

Reposted by Zhaoyang Wang

Alexander Terenin @avt.im · 07/05/2025

We’d like to announce next week’s virtual seminar on Bayesian Decision-making and Uncertainty! Sebastian Ament (Meta) Unexpected Improvements to Expected Improvement for Bayesian Optimization Sign-up link: gp-seminar-series.github.io

084

Reposted by Zhaoyang Wang

Anja Surina @anjasurina.bsky.social · 26/04/2025

Excited to share our latest work on EvoTune, a novel method integrating LLM-guided evolutionary search and reinforcement learning to accelerate the discovery of algorithms! 1/12🧵

12010

Reposted by Zhaoyang Wang

Professor Philip Torr, Oxford University @philiptorr.bsky.social · 24/04/2025

Cooperation on AI safety is necessary but also comes with potential risks. In our new paper, we identify technical AI safety areas that present comparatively lower security concerns, making them more suitable for international cooperation—even between geopolitical rivals. lnkd.in/e_FzCFW9

lnkd.in

This link will take you to a page that’s not on LinkedIn

011

Reposted by Zhaoyang Wang

Jeff Clune @jeffclune.com · 25/04/2025

Very excited for this keynote by @_rockt! Awesome to see open-endedness go from a niche (😉) area to a keynote at #ICLR ! 🌱🌿🌳🌲🍀🌍✨ 📈 🧬🧪 cc @joelbot3000.bsky.social @kennethstanley.bsky.social

0101

Reposted by Zhaoyang Wang

Alexander Terenin @avt.im · 24/04/2025

Online learning is like magic

281

Reposted by Zhaoyang Wang

Arthur Gretton @arthurgretton.bsky.social · 22/04/2025

Optimality and Adaptivity of Deep Neural Features for Instrumental Variable Regression #ICLR25 openreview.net/forum?id=ReI... NNs ✨better than fixed-feature (kernel, sieve) when target has low spatial homogeneity, ✨more sample-efficient wrt Stage 1 Kim, @dimitrimeunier.bsky.social, Suzuki, Li

083

Reposted by Zhaoyang Wang

Sam Power @spmontecarlo.bsky.social · 21/04/2025

Keen to read this: arxiv.org/abs/2504.13322 'Foundations of locally-balanced Markov processes' - Samuel Livingstone, Giorgos Vasdekis, Giacomo Zanella

2254

Reposted by Zhaoyang Wang

Arthur Gretton @arthurgretton.bsky.social · 15/04/2025

Looking forward to this!

052

Reposted by Zhaoyang Wang

Neil Lawrence @lawrennd.bsky.social · 07/04/2025

Check out our AI Core Concepts materials! Online training resource for researchers from all disciplines who want to work with AI.

042

Reposted by Zhaoyang Wang

Alexander Terenin @avt.im · 07/04/2025

This week’s virtual seminar on Bayesian Decision-making and Uncertainty is happening now! Frank Hutter (University of Freiburg) Accurate predictions on small data (and time series) with the tabular foundation model TabPFN YouTube: www.youtube.com/watch?v=SOXK... gp-seminar-series.github.io

0112

Reposted by Zhaoyang Wang

Gen AI - the AI Hub in Generative Models @genaihub.bsky.social · 04/04/2025

Our hub is funding a post doc researcher to work on projects involving the evaluation and improvement of multi-modal foundation models. Apply by 9th April. Great chance to work with @neuralnoise.com and @edinburgh-uni.bsky.social colleagues! elxw.fa.em3.oraclecloud.com/hcmUI/Candid...

elxw.fa.em3.oraclecloud.com

Research Associate

The School of Informatics, University of Edinburgh invites applications for a Post-Doctoral Research Associate in Natural Language Processing and Machine Learning to work with Dr Pasquale Minervini in...

042

Reposted by Zhaoyang Wang

Royal Statistical Society @royalstatsoc.bsky.social · 04/04/2025

Our new journal for #DataScience and #AI is open for submissions: academic.oup.com/rssdat?login... Back at its launch at our conference last September, editor in chief @lawrennd.bsky.social discussed how it fits with our mission and why its needed to confront current societal challenges

youtube.com

Launching our new journal for data science and artificial intelligence

YouTube video by RoyalStatSoc

0107

Reposted by Zhaoyang Wang

Maike Osborne @maosbot.bsky.social · 01/04/2025

Check out our workshop on Technical AI Governance! Submissions are due 7th of May.

072

Reposted by Zhaoyang Wang

Jes Frellsen @jesfrellsen.bsky.social · 31/03/2025

Now, the fantastic @jmtomczak.bsky.social is telling us about probabilistic circuits at GeMSS/Statlearn 2025 @gemssai.bsky.social

0166

Reposted by Zhaoyang Wang

ai@cam @ai.cam.ac.uk · 28/03/2025

Brilliant discussion last night from @dianecoyle1859.bsky.social @bennettinstitute.cam.ac.uk and @lawrennd.bsky.social on all things #AI for the @cambridgefestival.bsky.social

152

Reposted by Zhaoyang Wang

Gen AI - the AI Hub in Generative Models @genaihub.bsky.social · 28/03/2025

We are recruiting a post doc to analyse and improve multimodal foundation models who will be based at the University of Edinburgh. Apply through the university's website by 8 April.

063

Reposted by Zhaoyang Wang

ai@cam @ai.cam.ac.uk · 27/03/2025

We're looking forward to hearing from @dianecoyle1859.bsky.social @bennettinstitute.cam.ac.uk @lawrennd.bsky.social tonight as they discuss Neil's recent book The Atomic Human and why our fears of being displaced by #AI might be misplaced. 💭 Doors open from 17.30! 🎟️ lu.ma/y9kw5wk3

183

Reposted by Zhaoyang Wang

Neil Lawrence @lawrennd.bsky.social · 27/03/2025

Looking forward to being grilled!

031

Reposted by Zhaoyang Wang

Arthur Gretton @arthurgretton.bsky.social · 04/03/2025

Video now public from the talk "Learning to act in noisy contexts using deep proxy learning" at the NeurIPS'24 Workshop on Causal Representation Learning! Video: neurips.cc/virtual/2024... Slides: www.gatsby.ucl.ac.uk/~gretton/cou...

185

Reposted by Zhaoyang Wang

Eugene Vinitsky 🍒 @eugenevinitsky.bsky.social · 18/03/2025

Was great to have a chance to talk about our work on self-play for zero-sum games / our work on self-play for human-level simulated driving (I promise they're connected) @nyudatascience.bsky.social!

2283

Reposted by Zhaoyang Wang

Dan Goodman @neural-reckoning.org · 17/03/2025

If you've finished your PhD in the last 4 years and your research could be considered as engineering, you might be interested in the RAEng research fellowship scheme which provides 5 years of funding to work in the UK. raeng.org.uk/research-fel...

raeng.org.uk

Research Fellowships

02815

Reposted by Zhaoyang Wang

RL Theory Virtual Seminars @rl-theory.bsky.social · 17/03/2025

There is a new RL theory seminar tomorrow ! Join us :)

053

Reposted by Zhaoyang Wang

Thiago Serra @thserra.bsky.social · 16/03/2025

Willem-Jan van Hoeve talks at the INFORMS Computing Society Conference @informs.bsky.social @icsinforms.bsky.social about a general-purpose decision diagram-based discrete optimization solver, motivated by dynamic programming not being regarded more as an algorithmic than as a modeling device. 🧵1/N

152

Reposted by Zhaoyang Wang

Bojan Tunguz @tunguz.bsky.social · 16/03/2025

This is what they took away from us.

31053

Reposted by Zhaoyang Wang

Christian A. Naesseth @canaesseth.bsky.social · 15/03/2025

Don't forget to submit your Fast-Track papers to #AABI2025 today! If you have an accepted paper in machine learning since AABI 2024 present it as a poster. CfP: approximateinference.org/call/ Looking forward to seeing everyone in Singapore on April 29, just after #ICLR2025!

approximateinference.org

Call for Papers

044

Reposted by Zhaoyang Wang

Michael Bronstein @mmbronstein.bsky.social · 13/03/2025

2151

Reposted by Zhaoyang Wang

Sam Power @spmontecarlo.bsky.social · 10/03/2025

In case it's of interest to anybody: I got in touch with YN himself to ask a bit about the motivations behind his fast gradient method. I attach here (the relevant part of) what I asked:

I email here with a small historical question about the genesis of the Fast Gradient Method (FGM). Today, I was having a conversation with some colleagues about the method, and two of them raised the separate points that
you were a student and colleague of B. Polyak, who had worked on the Heavy Ball Method, which is quite connected to the FGM, but
reading your presentation of the FGM (in e.g. your textbook), the derivations rather focus more on the function-side perspective, with estimating sequences, and the connection to the physical interpretation is not highlighted in the same way.
In view of these perspectives, I wanted to reach out to ask how your motivation and intuition for the method came about initially. For example, was it coming from physical intuition, building on the work of Boris and others, or was it rather more mathematical / algebraic in character, or perhaps some mixture of the two? I'd be very interested to hear how you think about things.

1163

Reposted by Zhaoyang Wang

Exeter College, Oxford @exeter.ox.ac.uk · 05/03/2025

Anyone else loving how much lighter it is in the evenings 😍🌼 #ExeterCollegeOxford #RadcliffeCamera #OxfordUniversity

072

Reposted by Zhaoyang Wang

ai@cam @ai.cam.ac.uk · 10/03/2025

Don't miss your chance to join our #CamFest panel discussion! Our brilliant speakers include @lawrennd.bsky.social, Jess Montgomery, Tom Smith & local gov officials Liz Watts & Robert Pollock. 🗣️ How can we harness the power of AI for public benefit? 📆 2 April 🎟️ Register here: lu.ma/3e3spf0s

042

Reposted by Zhaoyang Wang

Dmytro Mishkin @ducha-aiki.bsky.social · 10/03/2025

MUSt3R: Multi-view Network for Stereo 3D Reconstruction Yohann Cabon, Lucas Stoffl, Leonid Antsfeld, Gabriela Csurka, Boris Chidlovskii, Jerome Revaud, @vincentleroy.bsky.social tl;dr: DUST3R for visual odometry - "just concatenate" (c) Amazing arch simplicity arxiv.org/pdf/2503.01661 #CVPR2025

0203

Reposted by Zhaoyang Wang

Maike Osborne @maosbot.bsky.social · 10/03/2025

CVPR, AAMAS and ICRA are all going to be held in the US this year. Is it fair to ask vulnerable students to travel to the US right now?

2322

Reposted by Zhaoyang Wang

Sam Power @spmontecarlo.bsky.social · 09/03/2025

oh cool news in the red there!

27212

Reposted by Zhaoyang Wang

Pascal @pvfua.bsky.social · 09/03/2025

A busy 24 hours around the @iccv.bsky.social deadline: Pizza night at the lab on Friday, 15 hours to ICCV deadline; finalizing papers on Saturday morning, 10 mns to deadline; our son’s civil wedding, 4 hours after deadline; wedding dinner, 9 hours after deadline.

2262

Reposted by Zhaoyang Wang

Maike Osborne @maosbot.bsky.social · 08/03/2025

I am 100% serious when I say that one of the most important skills for a researcher is an exceptional tolerance for feeling stupid. As a researcher, you really have to lean into feeling stupid.

517019

Reposted by Zhaoyang Wang

Alexander Terenin @avt.im · 21/02/2025

New preprint! This is a hardcore technical paper on Thompson sampling - as a strategy for the so-called online learning game. I think it's one of the most long-term important things I have ever worked on due to what it makes possible. That needs explaining: thread below! arxiv.org/abs/2502.14790

arxiv.org

An Adversarial Analysis of Thompson Sampling for Full-information Online Learning: from Finite to Infinite Action Spaces

We develop an analysis of Thompson sampling for online learning under full feedback - also known as prediction with expert advice - where the learner's prior is defined over the space of an adversary'...

2388

Reposted by Zhaoyang Wang

Andrew Gordon Wilson @andrewgwils.bsky.social · 28/02/2025

I had a great time talking with @anilananth.bsky.social as part of the Simons Institute Polylogues. We cover universal learning, generalization phenomena, how transformers are both surprisingly general but also limited, and the difference between statistics and ML! www.youtube.com/watch?v=Aja0...

youtube.com

Andrew Gordon Wilson | Polylogues

YouTube video by Simons Institute

082

Reposted by Zhaoyang Wang

Christian Wolf @chriswolfvision.bsky.social · 07/03/2025

#ICCV2025 crunch time picture, as requested! Our papers are almost ready, we will not need to go the last mile I think.

2292