A neural representation of continuous space using fractional binding

Code accompanying the paper presented at CogSci 2019 in Montreal

Brent Komer, Terrence C. Stewart, Aaron R. Voelker, and Chris Eliasmith. (2019) A neural representation of continuous space using fractional binding. Proceedings of the 41st Annual Meeting of the Cognitive Science Society.

Reproducing results

Figure 1, 2a, and 3 are created with example_figures.py

Figure 2b and 2c are created with exp_query_example.py

Data for Figure 4 is created by cap_acc_non_neural.py and cap_acc_neural.py for the spiking neural network results. Once the data is generated, it can be plotted with plot_capacity_and_accuracy.py

Data for Table 2 can be created by exp_neural.py and exp_non_neural.py. The output can be plotted with plot_exp_results.py. Note that the neural experiments can take a while to run on a standard machine with the default parameters.