Decoding analyses of the confidence dataset via linear support vector machine, random forest classifier and recurrent neural network models.

System Information

• Platform: Linux-3.10.0-514.el7.x86_64-x86_64-with-centos-7.3.1611-Core
• Python: 3.6.3 |Anaconda, Inc.| (default, Nov 20 2017, 20:41:42) [GCC 7.2.0]
• CPU: x86_64: 16 cores
• numpy: 1.16.4 {blas=mkl_rt, lapack=mkl_rt}
• scipy: 1.3.1
• matplotlib: 3.1.3 {backend=agg}
• seaborn: 0.11.1
• sklearn: 0.23.2
• pandas: 1.0.1
• tensorflow: 2.0.0
• pytorch: 1.7.1
• R: 4.0.3 # for 3-way repeated measure ANOVAs

Linear SVM

source: https://image.slideserve.com/867897/linear-support-vector-machine-svm-l.jpg

Random forest classifier

source: https://cdn-images-1.medium.com/max/1600/1*i0o8mjFfCn-uD79-F1Cqkw.png

RNN model - as an alternative model, but we do not perform model selection. An RNN model contains such prior: there exists temporal relationship between the features from consective time points and adding these relationships to the model would benefit the decoding.

Results - confidence

Decoding scores (within domain)

Using 7 trials back	Split to past and recent

Feature attribution (within domain)

Using 7 trials back - SVM	Using 7 trials back - RF

Decoding scores (cross domain)

Using 7 trials back	Split to past and recent

Feature attribution (cross domain)

Using 7 trials back - SVM	Using 7 trials back - RF