The Cognitive Labs Blog
It is not new the idea of making a blog so, why now? What does it mean? And most important, what is this about?
Six months ago, we released our new initiative to contribute to the open-source community by creating the Ericsson Cognitive Labs. We believed that since we are working in cutting-edge technology, such as Graph Neural Networks or Explainability, our research can benefit the community, having the possibility of applying it in fields such as medicine, autonomous driving, or drug discovery. The idea was bringing these ideas closer to the community and contributing to the open-source ecosystem to give back to the community. However, we have noted that sometimes academic papers, even if they bring an implementation with them, only reach a small audience.
To solve that, we have decided to bring a new idea to the table and make a technical blog. Many of you will know for sure the Ericsson Blog, where Ericsson projects are announced and discussed, but here our objective is completely different. As a result, our goal is to make our research more readily available, reaching a larger audience than a paper alone, and creating articles about specific implementations (sometimes with significant efficiency savings) that might not be appropriate for a journal or conference.
More specifically, we plan to add content in the following areas:
-
Bring AI research closer. There may be areas that are not as popular as the famous Large Language Models (LLMs), and one can find only little information about them. Moreover, these areas are usually only known in the academic world, with almost no information available outside these types of circles. Our first objective is to provide content about these AI topics, bringing them closer to the public so they can be used by anyone.
-
Deep dive into the software part of our research. We have the feeling that we always share the results of our experiments, which are extensively discussed in our papers, but never about the implementation or how to make research efficient. Even when sometimes we believe that the most interesting part of a problem is how to make it efficient and scalable.
-
Discussion about new technologies and methodologies in the AI industry. We also believe that it may be helpful to be fully open and transparent about what technologies we use, starting new discussions about the possible tech stack and how the industry is evolving.
What we have explained is not new, and we have been inspired by amazing companies such as Google DeepMind, Meta FAIR or Modular, and we want to take this idea one level further. Therefore, this will be not only to talk briefly about our projects or new trends but also a new channel to be able to share parts of our research that do not fit into more classic channels, such as conferences or journals. We are happy to share and contribute to this amazing industry. Stay tuned for more!