Group Meetings

Parenting Behavior and Long-Term Outcomes
Robyn Metcalfe

This meeting will outline an upcoming cross-disicplinary project between Computer Science and Counseling Psychology. As a part of a Counseling Psychology research project called Project Alliance, parent-child dyads from Portland Oregon have been tracked since middle school. In addition to extensive survey data, we have videos of parent-child interactions and audio recordings of parents and children discussing their relationship. The kids are now in early adulthood, meaning we know what happened to them! The proposed project will compare sentiment analysis results with the hand-coded parent-child interactions and use both questionnaire data and transcripts to predict 1. risk behavior 2. mental health outcomes and 3. arrest records.

Explaining Neural Networks
Isaac Ahern

Given enough data, deep neural nets are able to learn strong classifiers across various domains. However, in order to effectively use such a model for practical purposes, it's important to be able to explain what features are used for making predictions, both to establish user trust in the model, as well as to validate that it's learning useful and correct relationships. I will discuss a recent model-agnostic approach for prediction explanation by Ribeiro et. al. (2016), as well as the use of domain ontologies in order to improve both model predictions and explanations.

Regular Group Meeting on Feburary 7th
Gongming Wang

We will have our regular group meeting at Room 260 of Deschutes Hall. Gongming Wang will be doing a session on the topic 'Deep belief network with partial least square regression'.