Example Topics

• When to use what activation function in backprop.
• Second-order extensions of backpropagation.
• Applying nonlinear optimization techniques to training networks.
• Recurrent extensions of backpropagation.

• Hebbian learning and self-organization.
• Neuron as a principal component analyzer (Ch 8).
• Recent developments in self-organizing feature maps.
• Mathematical analysis of self-organizing feature maps.
• Feature extraction with neural networks.

• Theory of Radial Basis Function networks (Ch 5).
• Information-theoretic learning algorithms (Ch 10).
• Neural network learning vs. regression algorithms.
• Developing networks with genetic algorithms.
• Support vector machines (Ch 6).
• Learning algorithms that grow and delete units.
• Learning theory.

• Nonlinear dynamics and chaos in neural nets (Ch 14).
• Neural information processing based on
  • neural oscillators.
  • population oscillations.
  • synchronized firing.

• Neural networks and fuzzy systems.
• Modular neural networks.
• Committees of neural networks (Ch 7).

• Policy iteration methods of reinforcement learning.
• Coping with hidden states in reinforcement learning.
• The theory of reinforcement learning.
• Methods for evolving neural network topologies.
• Combinations of evolution and learning.
• Evolving neural networks for robot control.
• Evolving neural networks for game playing.

• Comparing neural nets and
  • statistical methods.
  • machine learning.
  • symbolic reasoning.
  • symbolic knowledge representation.
  • mathematical optimization.

• Super-Turing neural nets.
• Performance evaluation of neural nets.
• Benchmarking for learning algorithms.

• Learning and representing structure.
• Localist models in cognitive science.
• Marker passing models in cognitive science.
• Representing high-level knowledge in neural nets.
• Symbolic and sequential reasoning in neural nets.
• The connectionist -- symbolic debate.

• Extracting knowledge from a network by
  • clustering methods.
  • principal component analysis.
  • eyeballing.
  • magic.

• Neural network models of
  • low-level vision.
  • high-level vision.
  • episodic memory.
  • hippocampus.
  • human category learning.
  • aphasia, schizophrenia etc.
  • low-level neural processes.
  • the lexical system.
  • consciousness.

• Neural network applications on
  • speech recognition.
  • hand-written character recognition.
  • financial forecasting.
  • time-series analysis.
  • control.
  • expert systems.
  • natural language understanding.
  • machine translation.
  • game playing.
  • optimization.
  • performance optimization in computer systems.
  • fault diagnosis in computer systems.

• Implementation of neural nets in
  • optical hardware.
  • electronic hardware.
  • molecular hardware.
  • magnetic spin chip.
  • wetware cultures connected with chips.