Tom Viering

@FrankRHutter Cool! I was looking forward to reading this since my visit to Freiburg! I am particular fascinated by the generative abilities... In March I will give a lecture about PFNs and this will be included for sure :).

The data science revolution is getting closer. TabPFN v2 is published in Nature: nature.com/articles/s4158… On tabular classification with up to 10k data points & 500 features, in 2.8s TabPFN on average outperforms all other methods, even when tuning them for up to 4 hours🧵1/19

I'm hiring for a fully funded #PhD position in #ComputerVision on 'Detailed Video Understanding' at @LIACS @UniLeiden. Apply before 22nd June. More info👇 universiteitleiden.nl/en/vacancies/2…

The deadline for @NeurIPSConf 2024 has passed, but it doesn't mean we don't work hard to have another fantastic conference! Right now, we are in need for REVIEWERS. If you know someone suitable for this role, please fill in this form: forms.gle/W3Yrdqxu4jE11Q…

🚀 We're looking for a PhD student to join our team at @tudelft! Deadline for applications is January 5. Please share, and feel free to contact me if you have any questions.

@AliciaCurth @Jeffaresalan Great work! Excellent read!

Every StatML intro class covers complexity-error U-curves, so @Jeffaresalan & I asked ourselves whether the info from these classes is enough to explain double descent too? Our #NeurIPS23 paper does a roundtrip of The Elements of Statistical Learning and answers “Yes”! Long🧵1/n

@napoperez1998 Soro is right, please apply! We have several students from Latin America in our faculty.

@TViering Open to Latin America Students?, i am really interested

I am hiring a PhD student. The topic will be related to deep meta-learning on learning curves. You can apply here: tudelft.nl/over-tu-delft/…

@Burevestnik0829 Hi! If you are from TU Delft I have BSc or MSc projects. If you are from abroad but Europe, we can look at Erasmus. Other options also possible. DM me for more info :)

@TViering Any ug research assistant positions available 🥲

I will soon open up a PhD position for deep learning on learning curves. See here: tomviering.nl/vacancy.html Challenge: Developing deep meta-learning algorithms to model and understand learning curve patterns in machine learning. Impact: Faster, better, more cost-efficient ML.

And one more new MOOC, by Hongrui Wang, Amira lnouty, Luca laurenti, Wendelin Bohmer and @TViering. Covers AI concepts of clustering, dimensionality reduction, and introduces you to reinforcement and deep learning. Starts May 16th, so enroll now! online-learning.tudelft.nl/courses/ai-ski…

New MOOC by @TViering and Hanne Kekkonen: learn the basics of (supervised) machine learning, e.g. various classification and regression machine learning algorithms to solve real-life problems with scikit-learn in Python. Starts May 16th, so enroll now! online-learning.tudelft.nl/courses/ai-ski…

@cwolferesearch Exactly in line with Halevy, A., Norvig, P., & Pereira, F. (2009). The unreasonable effectiveness of data. IEEE intelligent systems, 24(2), 8-12.

"The bitter lesson of machine learning research argues that general methods that can leverage additional computation ultimately win out against methods that rely on human expertise." arxiv.org/abs/1910.10683

📢 #PhDposition! I have a vacancy @tudelft dealing with generating digital humans in social contexts. If you're interested in working at the intersection of multimodal ML, social cognition, and CV/graphics, this may be for you! linkedin.com/jobs/view/3537… #GenerativeAI #vacancy

@rjbruin @CVPR I think it makes sense. The discussion should start at the beginning of the discussion period, not beforehand. This ensures all reviewers can take part in the discussion equally.

@rjbruin @DrGroftehauge I think it may only be possible if your network is regularised (l2,l1, dropout) otherwise it cannot be the case due to a symmetry argument.

We love learning curves! This is why we created #LCDB (github.com/fmohr/lcdb), a database with API to LCs of 20 sklearn classifiers for 219 datasets, including pred. vectors for train/val/test folds. Check out @TViering's talk on the associated paper this Thursday at @ECMLPKDD.

@rjbruin @MatthiasBethge @bethgelab Enjoy!!! Be sure to also enjoy the surroundings ;)

@de_JQK @thegautamkamath @shortstein @icmlconf Thanks, will check!

@TViering @thegautamkamath @shortstein @icmlconf Thanks. Sorry but I do not have a blog. You could check the other replies under this thread where I mentioned som references. Hope it helps.

@de_JQK @thegautamkamath @shortstein @icmlconf Could you write a blog post on this? Or is there any paper reference on this? I would be very interested to know more...

@thegautamkamath @shortstein @icmlconf 1/n For LTH, they use very small learning rate to weaken the performance of the baseline of original dense model, and if you use unstructured sparsity to prune the network, the initialization condition derived by Kaiming should be carefully reexamined.