Kioxia Corporation, the global leader in storage solutions, has developed an image classification system based on Memory-Centric AI, an AI technology that leverages high-capacity storage. The system classifies images using a neural network related to knowledge stored in external high-performance memory; this avoids “catastrophic forgetting”, one of the main challenges of neural networks, and allows knowledge to be added or updated without losing current knowledge. This technology was presented on October 25th at the oral session of the European Conference on Computer Vision 2022 (ECCV 2022) in Tel Aviv, one of the top conferences in the field of computer vision[1].
This press release is multimedia. Check out the full release here: https://www.businesswire.com/news/home/20221101006375/en/
Image classification with high performance storage (Graphic: Business Wire)
In traditional AI techniques, neural networks are trained to acquire knowledge by updating parameters called “weights”. Once fully trained, in order to acquire new knowledge, a neural network must either be retrained from scratch or refined with new data. The former requires enormous amounts of time and consumes significant energy costs, while the latter requires updating parameters and faces the catastrophic forgetting problem of losing knowledge acquired in the past, leading to degradation in classification accuracy.
To address the cost and accuracy issues in neural network-based imagery…
