The LRZ has published an interview with Felix on the SWIM method developed by our group. This method allows for fast training of neural networks and drastically reduces energy consumption without sacrificing accuracy. See more details at… [weiterlesen]

We are happy to announce our workshop “Linear Operators in AI”.  The worlds of scientific computing and artificial intelligence are mostly developing in parallel. While "classical" scientific methods have strong mathematical foundations, they come with practical limitations, for example, in higher… [weiterlesen]

TUM has published an article on SWIM method developed by our group. This method allows for fast training of neural networks and drastically reduces energy consumption without sacrificing accuracy. See more details at… [weiterlesen]

Our new pre-print “Gradient-free training of recurrent neural networks” is out! Check it out at https://arxiv.org/abs/2410.23467. [weiterlesen]