Neural Network Models for Temporal Order Learning and Toad Habituation

Deliang Wang's dissertation from the University of Southern California presents formal and biological neural network models for learning. The first part develops a model for learning, recognizing, and reproducing complex temporal sequences using an attentional learning rule. The second part presents a computational model of visual pattern discrimination and synaptic plasticity in toads, validated by simulations against experimental data.

Use Cases

• Modeling short-term memory via mutually inhibiting neural units based on the described network architecture.
• Simulating acquisition and forgetting in synaptic plasticity based on the described interaction of two dynamic processes.
• Recognizing hierarchically organized temporal structures based on the described feedforward, chunking-like model layers.
• Reproducing complex temporal sequences while maintaining temporal course based on the described two-layer reciprocal connection architecture.

Strengths

• Model validation is described, with simulations reproducing experimental data 'remarkably well'.
• The work yields a range of experimental predictions concerning learning and pattern discrimination properties.
• Initial model testing experiments have validated certain predictions and led to new behavioral findings.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

Deliang Wang, University of Southern California

Collection Method

Theoretical and computational modeling, with large-scale computer simulations.