Share it

In recent years, artificial intelligence (AI) algorithms have vastly improved and are now used much more than just in the IT industry. Today they are to be found in areas including finance, medicine, art and renewable energy.
Specifically in computer vision (CV), major decisions are constantly delegated to these algorithms (medical diagnostics, autonomous cars, etc.), so it is crucial that these decisions are fair and explainable. In this context, we have the moral responsibility as a society to shift away from the concept of the “black box” in which we cannot explain the relationship between inputs and their outputs and move towards a point where any decision made by an algorithm can be interpreted.
In this Masterclass we will introduce the need for and the concept of explainability in AI to address the specific existing techniques for explainability in computer vision (focusing on the image classification task) and conclude by setting out use cases in which the interpretation of computer vision models is crucial.

Addressed to:

  • People interested in artificial intelligence and computer vision
  • Machine learning developers

Programa

  • What explainability in artificial intelligence is and why it is important
  • Explainability in computer vision models
  • Categories of techniques for explaining computer vision models
  • Use cases

Masterclass taught by NTT DATA (CIDAI core partner)

Taught by:

  • Stefano Masneri, Technical Manager in the Artificial Intelligence Knowledge Area at NTT Data. Stefano has 15 years of experience in image processing, computer vision and augmented reality and has worked as a researcher and computer vision engineer in Italy, Germany and Spain.
    Now at NTT Data, he is engaged in projects that help clients use AI to enhance their processes and generate added value
    .
  • Mario Mesas Rodriguez, Data Scientist at NTT Data Center of Excellence Mario is a Data Scientist at NTT DATA’s AI Center of Excellence. He has experience in applying computer vision techniques (object detection, instance segmentation) to multiple use cases from conception to production. With a particular interest in Edge AI and industrial applications of computer vision, at NTT Data he seeks to add knowledge and expertise to generate a differential product.