本文へジャンプ

  • HOME
  • 講演会情報

講演会情報

Machine Intelligence Project at IBM Research
講演者:Ahmet S. Ozcan(IBM Almaden Research)
平成30年7月20日(金)15時〜16時30分 理学部7号館1階102講義室
First I will give a high level overview of IBM Research and the activities at the Almaden Research Center in California. Then my talk will cover technical highlights from our latest work on Working Memory, which is summarized below:

Memory-augmented neural networks are a relatively new area of research. Augmenting neural nets with explicit memories give them remarkable abilities such as learning algorithms, task generalization and higher level abstraction. Recently, we designed a new model called Differentiable Working Memory (DWM) in order to specifically emulate human working memory. As it shows the same functional characteristics as working memory, it robustly learns psychology inspired tasks and converges faster than comparable state-of-the-art models. Moreover, the DWM model successfully generalizes to sequences two orders of magnitude longer than the ones used in training. Our in-depth analysis shows that the behavior of DWM is interpretable and that it learns to have fine control over memory, allowing it to retain, ignore or forget information based on its relevance.