Journal Club

Spring 2020: Machine Learning

BMERC is sponsoring a student-led cross-disciplinary journal club focused on machine learning, with particular applications in biomedical and drug discovery research. Over the past years machine learning has revolutionized the field of biomedical research. The goal of this journal club series is to introduce fundamental concepts of machine learning to life scientists and engineers, and thus encourage cross-disciplinary conversations and promote application of cutting edge machine learning methods in future research. You can find Jupyter Notebook sessions and Power Point slides on our dedicated github repository. The schedule of sessions is listed below:

Due to the recent developments related to the coronavirus (COVID-19) the first journal club of the semester, originally scheduled for March 18th, will be POSTPONED to a later date (TBD).

TBD

Paper: Machine learning to design integral membrane channelrhodopsins for efficient eukaryotic expression and plasma membrane localization

Topic: Applications of ML in protein engineering

Tutorial: N/A

Description:  We will cover a specific application of machine learning in protein engineering. Discussion of the selected paper will be led by BME student Jeffrey McMahan of the Ngo lab.

Past Meetings

— September 18, 2019 —

Paper: A Very Brief Introduction to Machine Learning…

Topic: Supervised Learning

Tutorial: Scikit-learn: SVM, Random Forest Examples

Description: We will explore the fundamentals of supervised learning using the selected reference paper. After a 20 to 30 minute presentation, a 5-10 min Q&A session will follow. Then, we will have a 20-30 min hands on session on Jupyter where we will use readily available data from scikit-learn to classify using SVM, Random Forest or other supervised ML method of participants’ choosing.

— October 23, 2019 —

Paper: A Very Brief Introduction to Machine Learning…

Topic: Unsupervised Learning

Tutorial: Scikit-learn: Clustering (K-means) and Classification (Nearest Neighbor)

Description: We will explore the fundamentals of unsupervised learning using the selected reference paper. After a 20 to 30 minute presentation, a 5-10 min Q&A session will follow. Then, we will have a 20-30 min hands on session on Jupyter where we will use readily available data from scikit-learn to cluster or classify using any unsupervised ML tool of the participant’s choice.

— November 20, 2019 —

Paper: TBD

Topic: Assessment of ML/DL Models

Tutorial: ROC Curve, Specificity, Selectivity, Accuracy, etc.

Description: We will explore the most common ways of evaluating machine learning models built in the first two sessions. Using the reference paper (TBD) and scikit-learn Model Evaluation, we will have a short presentation and Q&A session followed by a longer (30-40 min) hands-on session.

— December 11, 2019 —

Paper: TBD

Topic: BME/Chemistry Application of ML

Tutorial: N/A

Description: On the last day of ML journal club for the semester, we will have an application focused session. A volunteer participant will put together a 20-30 min presentation followed by either a hands-on session (if the algorithm/ML tool from the paper is available) or a discussion session where participants share how the algorithm/ML tool can be used in their research field.