Echo State Networks for Time-Series Data
Echo State Networks for Time-Series Data

Abstract: 

In this session, participants will be introduced to Echo State Networks (a type of recurrent neural network) including theory, key parameters in implementation and practical considerations. Participants will have the opportunity to use a publicly available Echo State Network implementation on open data. Additional results will be shown based on a highly customized implementation.
The tutorial will use python, jupyter notebooks and the EchoTorch python module (https://github.com/nschaetti/EchoTorch).
Participants will come away with a basic understanding of Echo State Networks, how to use the EchoTorch python module, the impact key parameters have on algorithm performance and potential application areas.

Bio: 

Teal Guidici is a Senior Machine Intelligence Scientist at Draper where she uses statistical techniques and machine learning algorithms to develop creative solutions for interesting data-driven problems in areas including biomedicine, finance, and remote sensing. Prior to Draper, she did graduate work creating new methods to analyze patterns of co-variation in complex datasets and applied these methods applied to high throughput metabolomics data. She has additional experience in survey design and data analysis in consumer marketing research. Dr. Guidici has a B.S. in Theoretical Mathematics from MIT, a M.S. in Bioinformatics and a Ph.D. in Statistics from the University of Michigan.