Stefan Haufe

Explainable AI techniques have been proposed as tools to "debug" machine learning models. In a popular example, authors discovered a photographer's watermark when applying XAI in an ima... Explainable AI techniques have been proposed as tools to "debug" machine learning models. In a popular example, authors discovered a photographer's watermark when applying XAI in an image classificat...

XAI techniques have been proposed as tools to "debug" machine learning models by revealing what they have learned.

Challenging this narrative, we find that XAI attributions are strongly dominated by salient features independent of the role of these features for a model.

See: tinyurl.com/48m4axnv

I am happy to announce the successful PhD defense of now Dr. rer. nat. Rick Wilming at TU Berlin, which he passed with distinction (summa cum laude). In his thesis entitled "Explainable AI under... I am happy to announce the successful PhD defense of now Dr. rer. nat. Rick Wilming at TU Berlin, which he passed with distinction (summa cum laude). In his thesis entitled "Explainable AI under Scru...

I am happy to announce the successful PhD defense of now Dr. rer. nat. Rick Wilming at TU Berlin, which he passed with distinction (summa cum laude) with this thesis on "Explainable AI under Scrutiny: From Empirical Investigations to Analytical Insights".

See:

www.linkedin.com/posts/stefan...

Thank you for the great work @willenjoy.bsky.social ! An important step towards understanding and correcting bias in source localizations and functional connectivity due to source leakage.

Work conducted by Benny Clark @qailabs.bsky.social . Kudos!

I am delighted to share our NeurIPS paper "Correcting misinterpretations of additive models" [1,2] to be presented by Benedict Clark in San Diego. Attending last year's "Interpreta... I am delighted to share our NeurIPS paper "Correcting misinterpretations of additive models" [1,2] to be presented by Benedict Clark in San Diego. Attending last year's "Interpretable AI: Past, Prese...

See my LinkedIn posting for further details:

www.linkedin.com/posts/stefan...

I am delighted to share our NeurIPS paper "Correcting misinterpretations of additive models" [1, 2] to be presented by Benny Clark in San Diego.

[1] openreview.net/pdf?id=2ClM0...

[2] recorder-v3.slideslive.com?share=107448...

After having pointed out severe shortcomings of existing XAI paradigms and feature attribution methods in the past, I am happy to announce two novel methodological contributions, to be presented in this year's main track of the premier ML conference, NeurIPS.

See LinkedIn post: tinyurl.com/25vu5xsm

Very useful software not only for neuroscientists provided by @tsbinns.com . Bispectral analyses allow robust determination of transmission delays, phase-amplitude coupling and quantification of non-sinusoidality of oscillations. See also our methods preprints:

lnkd.in/gSsrqYpX
lnkd.in/gCaXDggK

A huge thanks to you for the fantastic package, @tsbinns.com !

Brain Quantum Imaging We are excited to announce the Brain Quantum Imaging conference on the occasion of the opening of the joint OPM-MEG centre of Physikalisch-Technische Bundesanstalt and Charité-Universitätsmedizin. Thi...

Fantastic workshop on OPM imaging coming up at PTB, organized by our colleagues from Charité and PTB: conferences.ptb.de/event/15/

📌

Probably the most insightful and most important paper on physiomarkers of subthalamic DBS in a long while. Bravo @julianneumann.bsky.social @tsbinns.com and team!!

Out at @natcomms.nature.com:
www.nature.com/articles/s41...

Two days before deadline, looking forward to @overleaf.com being up again..!

We have a PhD opening in Berlin on "Responsible Data Engineering", with a focus on data preparation for ML/AI systems.

This is a fully-funded position with salary level E13 at the newly founded DEEM Lab, as part of @bifold.berlin .

Details available at deem.berlin#jobs-2225

Thanks a again to all who made it a success and specifically to @willenjoy.bsky.social for putting the book together!

We are happy to announce that the Proceedings of #NeuralTraces2024 have appeared: doi.org/10.14279/dep...

Time to commemorate this great event, hosted by the PTB in Berlin, and to browse through the interesting works on M/EEG methods and clinical applications presented.

Dear Volodymyr Zelenskyy, dear Ukrainian friends, you are not alone.

Great news: Thomas‘ study on shared and distinct mechanisms of PD therapeutics out soon! Thanks for the important work and great collaboration over the last years, @tsbinns.com , @julianneumann.bsky.social .

Das ZEW hat ausgerechnet, welche Einkommensgruppen von den Wahlprogrammen der Parteien jeweils am stärksten profitieren: www.sueddeutsche.de/projekte/art...

Explainable AI needs formal notions of explanation correctness | alphaXiv The use of machine learning (ML) in critical domains such as medicine poses risks and requires regulation. One requirement is that decisions of ML systems in high-risk applications should be human-und...

Happy to share our perspective preprint on the state of explainable AI. XAI needs to shift from algorithm- to problem-driven development to make sure methods can serve meaningful purposes. We are looking forward to discuss, here or on alphaxiv.org/abs/2409.14590 .

#NeuralTraces2024 at PTB Berlin concluded after three intense days. It was great to meet this high-flying crowd of researchers and to bring advanced M/EEG processing methods and their clinical applications a bit closer together. Thanks to all who participated!

We are excited to be kicking off #NeuralTraces2024 tomorrow! Participants can keep up to date by checking in on neuraltraces2024.org as well as the Indico platform. Note that we also offer late and single-day on-site admissions!

On Thursday, don't miss out Sotirios Papadopoulos, who will present his work on motor imagery decoding using beta beta bursts at Neural Traces 2024! Neuraltraces2024.org

One day left to register for Neural Traces 2024!

neuraltraces2024.org/registration...

Two!

neuraltraces2024.org/registration...

Four more days to register for Neural Traces 2024 in Berlin!

General participant registration for Neural Traces 2024 is closing on March 8th! Secure your spot under t.co/uhWavR49ft .

We look forward to seeing you in Berlin to talk M/EEG.

A few more days to submit to Neural Traces 2024 - our Advanced M/EEG Methods and Clinical Applications Workshop in Berlin!

We are excited to see a lot of great work being submitted to Neural Traces 2024 (neuraltraces2024.org/submission.h...) but also realize that some authors may be under pressure to submit in time. Therefore, we extended the abstract deadline one last time to Friday Jan 19, 9:00am CET.