Time Series II

Time Series II

author

Dušan Fedorčák

Prerequisites

  • basic knowledge of programing in Python
  • high school level of mathematics
  • Knowledge of machine learning on the level of our course Time Series

Abstract

The course is intended for people who want to deepen their knowledge and experience in the field of time series analysis. This is a follow-up course of Time Series, which is focused mainly on other domains of time series and their specifics. We will particularly study sound data with a combination of recurrent and convolutional neural networks and the sentiment analysis task on textual data.

Outline

  • Advanced feature preprocessing
  • Classification/pattern detection in sound data
    • Short-time Fourier transformation
    • CNN for sound classification: spectrogram as image
    • LSTM & CNN combination
    • Example: Speech detection using neural networks
  • Time series prediction from textual features
    • Data preprocessing
      • Sentence & word tokenization
      • Model-based vs. dictionary-based word embeddings
      • Of-the-shelf pre-trained word-embeddings models
    • Pre-training with auxiliary classification tasks
    • Model combination & fine tune training
    • Example: Time series predictions using sentiment analysis of news data

Dates

 

If you wish to enroll in this course please contact us on info@mlcollege.com.