Research Scientist / Assistant Professor in Artificial Intelligence, Machine Learning and Robot Learning
Italy
Joined September 2009
- Tweets1.4K
- Following236
- Followers97
- Likes81
A useful test for an AI system is not just whether it works under the benchmark assumptions, but whether the result survives when the constraints come from outside the experiment.
Curious what @OneMileataTime or @garyleff think — is breakage just part of the loyalty design, or a real user pain?
Do not miss it
Excited to share that our paper, “Transparent Explainable Logic Layers,” co-authored with @MarcPlantevit, Celine Robardet, and @webrot, will be presented at ECAI24! 🧠📊
Check it out! 📑 doi.org/10.3233/FAIA240579
#XAI #ECAI2024
Roberto Capobianco retweeted
A nice write up on our XAI for DRL workshop at AAAI earlier this year with @larosabiagio @runnerdude97
@webrot @spideralessio @MProietti_
#AAAI2024 workshops round-up 4: eXplainable AI approaches for deep reinforcement learning, and responsible language models
aihub.org/2024/04/12/aaai202…
Roberto Capobianco retweeted
Heading to #AAAI2024 in a few hours!! If anyone is around drop me a message on Whova and we can meet there! ps: visit our workshop in the quoted message!
Heading to #AAAI2024 today with @runnerdude97 to meet up with @larosabiagio and others for the XAI for DRL workshop: xai4drl.github.io/. Let me know if you're around!
Roberto Capobianco retweeted
Heading to #AAAI2024 today with @runnerdude97 to meet up with @larosabiagio and others for the XAI for DRL workshop: xai4drl.github.io/. Let me know if you're around!
Roberto Capobianco retweeted
I and @leilanigilpin are going to present the poster of this work tomorrow morning (10:45am) at #NeurIPS2023 . Paper #1522. Come and chat together about it!
Interested in understanding what neurons of #DeepNeuralNetworks are able to recognize at different ranges of activations? Thrilled to announce that our #XAI paper has been accepted at #NeurIPS2023 (openreview.net/forum?id=51PL…) and it is going to be presented as a poster in 🧵1/5 >>
Roberto Capobianco retweeted
Current LLMs are trained on text data that would take 20,000 years for a human to read.
And still, they haven't learned that if A is the same as B, then B is the same as A.
Humans get a lot smarter than that with comparatively little training data.
Even corvids, parrots, dogs, and octopuses get smarter than that very, very quickly, with only 2 billion neurons and a few trillion "parameters."
Roberto Capobianco retweeted
We just extended the deadline to November 21st!
Please retweet! We extended the deadline for our #XAI workshop XAI4DRL@AAAI2024 to Nov 21st! xai4drl.github.io/ Any XAI paper is welcome, even if RL is not involved! Please check the CFP and FAQ to know more about it!
Come join our team for an exciting internship!
As our work in Game AI expands, so does our team. Sony AI is now hiring for a Reinforcement Learning Research Intern to join our global Game AI group. Join us in leveraging AI to develop new, enriching gaming experiences. ow.ly/e7BI50Q6aek
Roberto Capobianco retweeted
This should be a fun workshop. Thanks for organizing!
Please RT! Two weeks left to submit your work to our workshop on #eXplainableAI approaches for Deep #ReinforcementLearning #AAAI24, Vancouver (xai4drl.github.io) Deadline: Nov 15!
Roberto Capobianco retweeted
Please RT! Two weeks left to submit your work to our workshop on #eXplainableAI approaches for Deep #ReinforcementLearning #AAAI24, Vancouver (xai4drl.github.io) Deadline: Nov 15!
Roberto Capobianco retweeted
Amazing (informal) news: Our paper has been accepted at #NeurIPS2023 and our workshop has been accepted at #AAAI2024 . I will share more details in the next few days about both of them. Both XAI-based! I think it's a great way to end my PhD journey. >>
Roberto Capobianco retweeted
1/9 We are very happy to share that our new paper titled "State of the Art of Visual Analytics for Explainable Deep Learning" has just been published by Computer Graphics Forum.
Open Access paper: doi.org/10.1111/cgf.14733
Interactive explorable survey: aware-diag-sapienza.github.i…
Roberto Capobianco retweeted
Train a weight matrix to encode the backpropagation learning algorithm itself. Run it on the neural net itself. Meta-learn to improve it! Generalizes to datasets outside of the meta-training distribution. v4 2022 with @LouisKirschAI arxiv.org/abs/2012.14905
Roberto Capobianco retweeted
I'm happy to announce that our paper 📄"A self-interpretable module for deep image classification on small data" has been published today in the Applied Intelligence Journal (rdcu.be/cS4P5). 1/6 >>
Roberto Capobianco retweeted
Episode 4. The one with "Outracing champion Gran Turismo drivers with deep reinforcement learning" by @webrot @Florian_fuu @leilanigilpin @PeterStone_TX @heyitdeclan @csherstan @neonleonb @HKitano #GTSophy @SonyAI_global
🎙️🙌
rss.com/podcasts/the-weekly-…
👇📑
Roberto Capobianco retweeted
Gradient descent/ascent can be inefficient to find saddle point (Nash equilibrium) for min-max games, because of spiralling behaviour. en.wikipedia.org/wiki/Nash_e…
