The Information plane can be used to gain insight and theoretical understanding of neural networks.

Image:

Opening the black box of AI

Although Deep Neural Networks (DNNs) are at the core of most state–of–the art systems in computer vision, the theoretical understanding and explainability possibilities of such networks is still not at a satisfactory level.

To a large degree, our user partner’s applications involve imaging the unseen – the inside of the human body, the sea, and the surface of the earth seen from space independent of daylight and weather conditions. Impact of innovative technology for users depends on trust. A limitation of deep learning models is that there is no generally accepted solution for how to open the “black-box” of the deep network to provide explainable decisions which can be relied on to be trustworthy.

Visual Intelligence aims at developing deep learning models with built-in robustness to data domain shift that also offers a high-level explainability.

Video

Further reading

Detection and classification of fish species from acoustic data
March 1, 2021
We collaborate with the Institute of Marine Research (IMR) to develop models and applications to detect and classify fish from echosounders.
Detection of sea mammals from aerial imagery
December 18, 2020
Better solutions are needed to estimate the populations of sea mammals, such as breeding seals, from aerial images of the sea ice.
Deep learning and AI in the medical domain
January 19, 2021
Overcoming the challenges of limited training data in the medical domain and laying the fundamentals for explainability and reliability.