Science of Science Summer School (S4) Lectures

Introduction

These are the lecture slides, notebooks, datasets, and exercises for for the Science of Science Summer School (S4) 2021 (https://s4.scienceofscience.org).

S4 2021 is hosted by the School of Information Studies at Syracuse University and organized by Daniel Acuna and Stephen David.

Structure

Day 1: Introduction

• Presentation: Opening and introduction to science of science
• Teaching #1 Introduction to environment: Jupyterhub, Notebooks, GitHub repository
  • Activity #1: Login into the system, run a notebook, save, and submit through nbgrader
• Teaching #2: explore datasets (MAG sample), funding (Ying Ding), mentorship (Qing Ke), content (pubmed open access), images (pubmed open access)
  • Activity #2: simple computation of citations, funding across years, mentorship, text, and images
• Teaching #3: Introduction to Python: basic principles, loading libraries, debugging
  • Activity #3: run simple program on Python, load data into Pandas, and run a simple regression

Day 2: Machine learning and artificial intelligence

• Presentation: Overview of machine learning in science of science
• Teaching #1: Probability, statistics, learning, errors, functions
  • Activity #1: Different kinds of learning and functions
• Teaching #2: Model complexity and interpretability
  • Activity #2: show how to overfit, underfit, bias-variance tradeoff
• Teaching #3: unsupervised learning, semi-supervised learning
  • Activity #3: dimensionality reduction, NLP, reinforcement learning
• Teaching #4: TBA
  • Activity #4: TBA

Day 3: Network science

• Presentation: Overview of Network Science
• Teaching #1
  • Activity #1
• Teaching #2
  • Activity #2
• Teaching #3
  • Activity #3
• Teaching #4
  • Activity #4

Day 4: Deep learning

• Presentation: Lucy Wang from Allen Institute of Artificial Intelligence (AI2)
• Teaching #1: Neural networks (neurons and learning)
  • Activity #1: Try neural network playground
• Teaching #2: Models for temporal data (BiLSTM, Transformers, etc)
  • Activity #2: Citation worthiness prediction
• Teaching #3: Models for image analysis (CNN, ResNet)
  • Activity #3: image analysis, misleading graphs
• Teaching #4: Bias in AI (Lizhen's presentation)
  • Activity #4: Example of Bias in AI

Day 5: Causal inference

• Presentation by Jianxuan Liu (Syracuse University)
• Teaching #1: From correlation to causation - intuitive example
  • Activity #1: Try discovering whether there is causality
• Teaching #2: Methods for causal inference - theory of propensity score matching
  • Activity #2: Simple example using PSM with logistic regression
• Teaching #3: Difference in difference, regression discontinuity, matching
  • Activity #3: Example from the literature (Aaron Clauset, Dashun's paper)
• Teaching #4: Machine learning perspective (do-calculus), DAGs
  • Activity #4: Backdoor, transportability, etc