Damien Teney
@damienteney.bsky.social
Research Scientist @ Idiap Research Institute. @idiap.bsky.social
Adjunct lecturer @ Australian Institute for ML. @aimlofficial.bsky.social
Occasionally cycling across continents.
https://www.damienteney.info
Adjunct lecturer @ Australian Institute for ML. @aimlofficial.bsky.social
Occasionally cycling across continents.
https://www.damienteney.info
Academic Strava?🤓 It feels like an underrepresented group in my Strava feed!
July 26, 2025 at 10:10 AM
Academic Strava?🤓 It feels like an underrepresented group in my Strava feed!
It'd be nice to provide complete analyses (that you have precomputed) of existing papers, so we can see what kind of output the tool provides, without having to submit any of my own work.
July 24, 2025 at 8:24 AM
It'd be nice to provide complete analyses (that you have precomputed) of existing papers, so we can see what kind of output the tool provides, without having to submit any of my own work.
Dang it just never ends 😱
July 22, 2025 at 6:08 PM
Dang it just never ends 😱
In this setting, does the student (sometimes?) get better than the teacher? One hypothesis could be that the teacher, even if "less correct" than the GT, provides supervision that's easier to learn for another NN (the student). The optimization follows a less tortuous path & finds a better solution.
July 16, 2025 at 3:24 PM
In this setting, does the student (sometimes?) get better than the teacher? One hypothesis could be that the teacher, even if "less correct" than the GT, provides supervision that's easier to learn for another NN (the student). The optimization follows a less tortuous path & finds a better solution.
👍 I had my very first paper published at DAGM. It was a while ago but I remember it as a very welcoming conference.
July 7, 2025 at 7:56 PM
👍 I had my very first paper published at DAGM. It was a while ago but I remember it as a very welcoming conference.
🎯There's already a plethora of methods to handle distribution shifts: most gains may now simply be in better using them! Automatic selection looks promising, yet there's lots more to do. Interested? Come chat with us at ICML!
📄 arxiv.org/abs/2410.02735
💻 github.com/LiangzeJiang...
📄 arxiv.org/abs/2410.02735
💻 github.com/LiangzeJiang...
OOD-Chameleon: Is Algorithm Selection for OOD Generalization Learnable?
Out-of-distribution (OOD) generalization is challenging because distribution shifts come in many forms. Numerous algorithms exist to address specific settings, but choosing the right training algorith...
arxiv.org
July 7, 2025 at 4:51 PM
🎯There's already a plethora of methods to handle distribution shifts: most gains may now simply be in better using them! Automatic selection looks promising, yet there's lots more to do. Interested? Come chat with us at ICML!
📄 arxiv.org/abs/2410.02735
💻 github.com/LiangzeJiang...
📄 arxiv.org/abs/2410.02735
💻 github.com/LiangzeJiang...
🔎Last but not least: OOD-Chameleon shines new light on existing algorithms! We can interpret the selection process as a tree and get interpretable guidelines for choosing algorithms.
July 7, 2025 at 4:51 PM
🔎Last but not least: OOD-Chameleon shines new light on existing algorithms! We can interpret the selection process as a tree and get interpretable guidelines for choosing algorithms.
✅We test OOD-Chameleon on unseen datasets & shifts: it accurately predict suitable algorithms on synthetic, vision, and language tasks. The downstream models (trained with selected algorithms) consistently have lower error than with standard selection heuristics.
July 7, 2025 at 4:51 PM
✅We test OOD-Chameleon on unseen datasets & shifts: it accurately predict suitable algorithms on synthetic, vision, and language tasks. The downstream models (trained with selected algorithms) consistently have lower error than with standard selection heuristics.
🛠️To create our "dataset of datasets", we resample CelebA and CivilComments with constraints specifying diverse types/magnitudes of shifts. We also train small models with candidate algorithms to obtain their "ground truth performance" in each condition.
July 7, 2025 at 4:51 PM
🛠️To create our "dataset of datasets", we resample CelebA and CivilComments with constraints specifying diverse types/magnitudes of shifts. We also train small models with candidate algorithms to obtain their "ground truth performance" in each condition.
🦎We propose OOD-Chameleon as a proof-of-concept: an algorithm selector as a classifier (over candidate algorithms) trained on a "dataset of datasets" representing diverse shifts. The model learns which algorithms perform best in different conditions.
July 7, 2025 at 4:51 PM
🦎We propose OOD-Chameleon as a proof-of-concept: an algorithm selector as a classifier (over candidate algorithms) trained on a "dataset of datasets" representing diverse shifts. The model learns which algorithms perform best in different conditions.
💡We're aiming for an "auto-ML for distribution shifts". We conjecture that datasets have properties predictive of the suitability of various algorithms to handle dist. shifts: size/complexity of the data, magnitudes/types of shifts, etc.
July 7, 2025 at 4:51 PM
💡We're aiming for an "auto-ML for distribution shifts". We conjecture that datasets have properties predictive of the suitability of various algorithms to handle dist. shifts: size/complexity of the data, magnitudes/types of shifts, etc.
"Distribution shift" means many things: spurious correlations, covariate shift, label shift... with no one-size-fits-all! Many algorithms exist, 📊each for specific conditions. Could we automate the selection without trial-and-error❓
July 7, 2025 at 4:51 PM
"Distribution shift" means many things: spurious correlations, covariate shift, label shift... with no one-size-fits-all! Many algorithms exist, 📊each for specific conditions. Could we automate the selection without trial-and-error❓
On the contrary it's a discussion that needs bringing up inside the CV research community. They're not just a bunch of white dudes with evil intentions or industry pressure to build a surveillance state. As one example see the thoughts from one such researcher lucasb.eyer.be/snips/cv-eth...
Ethical considerations around Vision and Robotics
A rough outline on how I think about doing research in Computer Vision given the many possible unethical uses.
lucasb.eyer.be
June 30, 2025 at 4:25 PM
On the contrary it's a discussion that needs bringing up inside the CV research community. They're not just a bunch of white dudes with evil intentions or industry pressure to build a surveillance state. As one example see the thoughts from one such researcher lucasb.eyer.be/snips/cv-eth...
Brilliant! Some pushback I heard against mandatory reviewing was from people misunderstanding that a submission entails such a partnership with the rest of the community.
June 30, 2025 at 8:26 AM
Brilliant! Some pushback I heard against mandatory reviewing was from people misunderstanding that a submission entails such a partnership with the rest of the community.
Looks quiet! Best time of day 👌🏼
June 30, 2025 at 7:42 AM
Looks quiet! Best time of day 👌🏼
Good points. It's a difficult task to automate the classification of papers/patents. Would the tracking of hands/gestures for sign language interfaces count as surveillance here?
June 30, 2025 at 6:23 AM
Good points. It's a difficult task to automate the classification of papers/patents. Would the tracking of hands/gestures for sign language interfaces count as surveillance here?
Because it discredits and could silence an entire area of scientific inquiry. Inclusive access to technology would tremedously benefit from CV technologies, so surveillance-based commercialization is only one part of the conversation.
June 29, 2025 at 6:26 PM
Because it discredits and could silence an entire area of scientific inquiry. Inclusive access to technology would tremedously benefit from CV technologies, so surveillance-based commercialization is only one part of the conversation.
There's a healthy amount of skepticism to be had when reading any paper, whatever level of peer-review it got through. Just like I wouldn't blindly trust a CVPR or NeurIPS paper claiming to beat the SOTA, because the authors were motivated to do so. Happy to continue the discussion elsewhere.
June 29, 2025 at 6:07 PM
There's a healthy amount of skepticism to be had when reading any paper, whatever level of peer-review it got through. Just like I wouldn't blindly trust a CVPR or NeurIPS paper claiming to beat the SOTA, because the authors were motivated to do so. Happy to continue the discussion elsewhere.
I'd be an important topic worth addressing more deeply within the CV community! We'd need more data on dual use and on beneficial application of the technology as well.
June 29, 2025 at 6:00 PM
I'd be an important topic worth addressing more deeply within the CV community! We'd need more data on dual use and on beneficial application of the technology as well.
Indeed, when reading the scientific literature, even with any amount of scrutiny and peer review, you ultimately have to trust the authors that they did their due diligence and didn't cut corners to arrive at their desired conclusions.
June 29, 2025 at 5:56 PM
Indeed, when reading the scientific literature, even with any amount of scrutiny and peer review, you ultimately have to trust the authors that they did their due diligence and didn't cut corners to arrive at their desired conclusions.
I'm open to any finding, there's no place for feelings when interpreting data. But I'm not sure this was the case for the authors. It's difficult to trust this paper because it feels like a piece of activism rather than an unbiased scientific study.
June 29, 2025 at 5:43 PM
I'm open to any finding, there's no place for feelings when interpreting data. But I'm not sure this was the case for the authors. It's difficult to trust this paper because it feels like a piece of activism rather than an unbiased scientific study.
Good for you. I said *most*, and was just reflecting on typical members of the community from the CV venues studied in the paper.
June 29, 2025 at 5:06 PM
Good for you. I said *most*, and was just reflecting on typical members of the community from the CV venues studied in the paper.
I think it's to the detriment of the authors, bc such a biased activist message will be ignored by most of the very ppl (CV researchers) that the information could have an effect on. This should have been presented at a CV conference if they were hoping for actual impact. @abeba.bsky.social
June 29, 2025 at 4:24 PM
I think it's to the detriment of the authors, bc such a biased activist message will be ignored by most of the very ppl (CV researchers) that the information could have an effect on. This should have been presented at a CV conference if they were hoping for actual impact. @abeba.bsky.social