Roberto Santana

Very informative presentation on current challenges for #hydrogen #energy #economy by Prof. Rasul at i-COSTE conference in @MurdochUni. While current focus is on technological challenges in infrastructure, computational and mathematical modeling questions deserve attention.

@el_pais Ojalá el esfuerzo que dedicamos a preparar los proyectos de IA no caiga en saco roto, y el compromiso del Gobierno al lanzar la convocatoria se mantenga. No hay justificación para minar la credibilidad ni debilitar la confianza de los investigadores en @CienciaGob @DianaMorantR

“Estamos supermaltratados, pero esto es un nivel superior”, dice un catedrático que pidió una de las ayudas para investigación en IA que se convocaron en julio del año pasado. “Cuando un Gobierno decide promocionar algo y luego no lo hace, nos fastidia” #Echobox=1733478940" target="_blank" rel="nofollow noopener">elpais.com/tecnologia/202…

Come an join our team at BCBL in Spain! A few days to apply for an exciting PhD position to investigate perceptual awareness using psychophysics, fMRI and decoded neurofeedback. Apply here:bcbl.eu/en/join-us/job… Deadline: December 1, 2024 Pls retweet!

Conference ADMA-2024. Presenting our poster summarizing some results of the IA-TERM project on the application of AI techniques to thermal installations in buildings. @isg_ehu @txiomatika

@f_charton @bartzbeielstein Modeling promising solutions, sampling new ones, selecting the best, and/or applying local optimizers is at the core of estimation of distribution algorithms (EDAs), used for optimization. Models used include Bayesian/Markov networks, generator NNs (GANN,VAE) and simpler models.

Transformers for discrete optimisation problems 1- Train a model on candidate solutions 2- Use the model to generate more candidates 3- Improve the solutions with local search 4- Use the best candidates to fine tune the model 5- Iterate

Our paper analyzing the vulnerability of explainable machine learning models to adversarial attacks is now online: wires.onlinelibrary.wiley.com/doi/10.1002/wi…

Our paper on the study of the generalization of physics-informed neural networks is out (link.springer.com/article/10.100…). We found that overfitting and performing highly accurate training proved to be beneficial concepts for #PINNs in terms of their generalization capabilities.

Estimation of Distribution Algorithms are 30 years old now, and a @PPSN2024 workshop celebrates the occasion. Great, comprehensive talk by Pedro Larrañaga and interesting presentations.

new preprint "Domain Adaptation-Enhanced Searchlight: Enabling brain decoding from visual perception to mental imagery" with @ai_research @nadine_dijkstra @meganakpeters

📚The "Handbook of Evolutionary Machine Learning" is out now! Spanning 750+ pages and 26 chapters, it explores fundamental concepts, advanced topics, and real-world applications. 🔗 Explore the handbook: link.springer.com/book/10.1007/9…

0/. Brain is not ViT. we scored 70.8 in the Algonauts 2023 visual brain competition, w/o ensemble we can do 66.8 score. other teams (including me in the past) are struggling with 60. “memory” is the secret. paper: arxiv.org/abs/2308.01175 code&webpage: huzeyann.github.io/mem

Smooth #GECCO2023 presentations of two works on permutation-based optimization problems by members of @isg_ehu of @txiomatika

It looks like evolutionary algorithms will also get entangled with the unfulfilled promises of quantum computing, so while waiting for supremacy let's learn more about QC in the #GECCO2023 workshop.

An interesting variety of goals and perspectives in the papers of the #GECCO2023 workshop on Enhancing Generative Machine Learning with Evolutionary Computation. gecco-2023.sigevo.org/dl1237

Great turnout and 9 fantastic speakers at the Workshop on Landscape-Aware Heuristic Search @GeccoConf on this first morning of #GECCO2023

Couldn't make it to #IJCNN2023 in person? Join us online! Our workshop on Multimodal Synthetic Data for Deep Neural Networks is free for virtual attendees. Talks kick off at 8.30am tomorrow. Link to join: #success" target="_blank" rel="nofollow noopener">us02web.zoom.us/j/88975862664#… Full program: bit.ly/MSynD

Insightful comments and high-quality submission in the Competition on Reproducibility and Loss Landscape Mapping in Feed-forward Neural Networks organized as part of #ijcnn23.

Opening of #ijcnn23 conference on Neural Networks with excellent talk by Prof. Xin Yao on #multiobjective ensemble learning.

I'm excited to be organising a competition on "Reproducibility and Loss Landscape Mapping in Feed-forward Neural Networks" at IJCNN2023! Please circulate to anyone you think might be interested. Participation in person or online is possible. marcusgal.github.io/reproduce_nnex…