26/04/2023.
Diffusion models rival and can even surpass GANs on image synthesis, as they are able to generate more diverse outputs by having a better coverage of the data distribution and do not suffer from mode collapse and training instabilities of GANs.
25/04/2023.
Hungary’s oldest still operating medical and health sciences higher education institution, Semmelweis University, invited SZTAKI deputy director and iToBoS colleague Róbert Lovas to give a presentation about European Union-funded projects that foster medical research and innovation.
24/04/2023.
Transformers are deep learning architectures created to solve sequence-to-sequence tasks, such as language translation.
23/04/2023.
Just shy over a year ago, the Quantus toolkit v0.1.1 has been shared with the Machine Learning (ML) community as a pre-print on arXiv.org.
22/04/2023.
iToBoS was discussed in a presentation at the 19th EADO (European Association of Dermato Oncology) Congress, held at Rome, Italy, in April 2023.
21/04/2023.
Explainable Artificial Intelligence (XAI) can not only be employed to get an insight into the reasoning process of an Artificial Intelligence (AI) model.
19/04/2023.
The paper ‘Attention Is All You Need’ introduces transformers and the sequence-to-sequence architecture. This is a neural net that transforms a sequence of elements (for example, the sequence of words of a sentence) into another sequence.
18/04/2023.
The iToBoS 4th Project Management Board (PMB) meeting took place on 18/04/2023 throughout a video conference system, with the participation of 19 attendees, including the Project Coordinator (PC), Project Manager (PM), Innovation and Exploitation Manager (IEM), Dissemination and Communication Manager (DCM), Data Manager (DM) and WP leaders.
17/04/2023.
Rule-based eXplainable AI (XAI) methods, such as layer-wise relevance propagation (LRP) and DeepLift, provide large flexibility thanks to configurable rules, allowing AI practitioners to tailor the XAI method to the problem at hand.
14/04/2023.
As we saw in a previous post, some challenges caused in explaining neural network decisions can be overcome via canonization.