Home > About Fuji Electric > Research & Development > R&D Activities > Representative R&D Projects > Deep learning technology capable of inference result explanation

Deep learning technology capable of inference result explanation

  • Digital innovation Laboratory
  • Fundamental and cutting-edge technologies
  • Safety and Security
  • IoT
  • Automatic optimizing model construction
  • Inference results explanation by quantifying input/output relationships
  • Automatic model updating through close-quarters learning

Deep learning, a typical AI technology, is expected to be applied to highly accurate demand forecasting, as well as quality prediction for plants and so on. However, deep learning requires manual parameter (model construction) tuning and maintenance (model updating). Furthermore, as it is difficult to interpret inference results, applications in industrial fields, which require long-term operation and reliability, have thus far been limited.

In response to this, Fuji Electric has been working on the development of deep learning technology capable of automatic learning and inference result explanation.

We are aiming to expand applications for industrial fields through the adoption of this technology.

Deep learning technology capable of inference result explanation