#imbalanced data

3posts

  1. Reading Multiple Emotions Mixed in a Single Utterance from Speech
    Can multiple emotions mixed within a single Korean speech clip be detected at the same time? We compare seven models built on three types of speech features, and outline the best combination and the limitations of using it in practice.
  2. Can Acoustic Features of the Voice Tell Us About Cognitive Status?
    Can cognitive status be distinguished using only acoustic features extracted from short vocal recordings of 223 Korean speakers? This post summarizes a single paper covering model performance, important features, and the limitations posed by differences in age and education.
  3. Can We Spot NOx Spikes in Electric Arc Furnaces Before They Happen?
    Can a NOx surge in a ferroalloy electric furnace be detected in advance from process data alone? This post examines the performance and limitations of an LSTM autoencoder, trained only on normal operation, in picking out signals five minutes before a surge.

Keywords