Curious AI’s solutions are not limited by or bound to any existing platform or technology, whether RPA, ERP or databases. We brew our own.
Curious AI develops advanced AI algorithms and solutions. We aim to generalise (AGI) our machine learning technology components for a software solution capable of human-level knowledge work: a digital co-worker.
For Curious AI’s goal of delivering automated knowledge work solutions, there are 3 crucial R&D focus areas: learning, perceiving, and autonomy.
First, we are best known for our extensive work on semi-supervised machine learning, including the seminal paper on Ladder neural networks. Helsinki’s long tradition (see founder Dr Harri Valpola’s 25+ years in AI research) in unsupervised learning is a definite advantage. Second, Curious AI has worked for years on perception systems. Our research interests include machine attention, segmentation (with the published Tagger technology), and perceptual grouping. Third, Curious AI is a forerunner in the field of autonomy via our research in e.g. model-based reinforcement learning and model-predictive control. These advanced proprietary technologies we are not currently publishing, but rather taking into production in e.g. our digital co-worker solution for industrial process operators.
Curious AI performs cutting-edge academic and industrial research in various AI disciplines. Some of our recent research topics are model-predictive control, few-shot learning, and neurosymbolic representations. One aspect in our approach is often unsupervised machine learning, in which CAI’s core team and the Helsinki AI scene in general packs decades of experience. A sampling of recent publications can be found on the Materials page.