Abstract

This paper presents our contribution to ACM ICMI 2014 Mapping Personality Traits Challenge and Workshop. The proposed system utilizes Extreme Learning Machines (ELM) and Canonical Correlation Analysis (CCA) for modeling acoustic features. The ELM paradigm is proposed as a fast and accurate alternative to train Single Layer Feed-forward Networks (SLFN) and Support Vector Machines (SVM). Benefiting from the fast learning advantage of ELM, we carry out extensive tests on the data using moderate computational resources. We further investigate the suitability of a recently proposed feature selection approach to prune the acoustic features, as well as mean smoothing of predictions. In our study, Kernel ELM performed better than basic ELM. Though an average (6-fold cross-validation) Pearson's correlation of 0.642 is reached on the training and validation sets, the overall correlation obtained on the sequestered test set is very low. The results indicate the difficulty of the problem.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call