Cognitive Modeling

Recent Advances in Recurrent Neural Networks

Dozent Dr. Sebastian Otte
Zeit

Fr. 10 c.t. bis 12 Uhr
Umfang 3 LP
Beginn 18.10.2019
Ort Informatik, Sand 14, PC-Poolraum C 412
Ilias https://ovidius.uni-tuebingen.de/ilias3/goto.php?target=crs_2020915

Seminar topics

  • Introduction to Spiking Neural Networks
  • Recent Advances in Neuromorphic Hardware
  • FastGRNN
  • Hardware Implementations of LSTMs
  • Deep Echo State Networks
  • RNNs and Chaos
  • Temporal Convolution vs. RNNs
  • Backpropagation through time in Spiking Neural Networks
  • Backpropagation through time and the brain
  • Gated Orthogonal Recurrent Units
  • Tensorized LSTMs
  • h-detach: Modifying the LSTM Gradient Towards Better Optimization
  • Cortical microcircuits as gated-recurrent neural networks
  • Non-normal Recurrent Neural Networks
  • Reversible RNNs
  • State-Regularized Recurrent Neural Networks
  • Quantifying Robustness of Recurrent Neural Networks
  • Ordered Neurons: Integrating Tree Structures Into RNNs
  • Fast Interactive Object Annotation with Curve-GCN
  • Speech Synthesis from Neural Decoding of Spoken Sentence
  • Streaming End-To-End Speech Recognition for Mobile Devices
  • Solving Rubik's Cube with a Robot Hand
  • AlphaStar: Mastering the Real-Time Strategy Game StarCraft II