神經科學資料分析
Data Analysis in Neuroscience
| 節 | 週三 |
|---|---|
3 10:10–11:00 | 神經科學資料分析 YL839 2 節連堂 |
4 11:10–12:00 |
* 根據陽明交大上課時間表所列
This course is intended to provide an overview on the statistical, analytic, and computational tools that are commonly used in the research field of neuroscience. It will focus on characterizing the strength and applicability of various data analytical approaches in some more intuitive than formal ways. Through practical, simplified exercises, it aims to initiate students with tools that are likely to be useful for their future research in the field. Following the textbook the tentative outline is as following Statistical inference: What is the data telling us? Regression problems: Possible trends of the data. Classification problems: Asking questions and finding answers in the data. Model complexity and selection: Are we reading too much from the data? Clustering and density estimation: Characterize the “shapes” of the data. Dimensionality reduction: Cut out irrelevance, finding the most important. Linear time series analysis: Things that add up to the changing data. Nonlinear concepts in time series analysis: Chemistry between the causes of change. Time Series from a Nonlinear Dynamical Systems Perspective: The ways that moving things can mingle and what we will see as results. Demonstration and implementation will be done in the Python programming language. Prior experience in Python will be helpful but not required.
Required: effective English communication ability Helpful: basic calculus, linear algebra, and programming experience
Lectures will follow the main textbook. Homework will generally be assigned at the conclusion of each textbook chapter. You will have a week's time to complete your homework. Homework should be submitted in a single ipynb or a single PDF file through the E3 digital learning platform (E3 數位教學平台).
Homework (90%) Take-home final (10%)
| 週次 | 主題 |
|---|---|
| 第 1 週 | Course introduction Statistical models Model-based analysis Parameter estimation |
| 第 2 週 | Optimizations: gradient descent, expectation maximization Hypothesis testing |
| 第 3 週 | Multiple linear regression General linear model Multivariate regression Canonical correlation analysis |
| 第 4 週 | Ridge & LASSO regression Local linear regression Basis expansions & splines k-nearest neighbor method Artificial neural networks and nonlinear regression |
| 第 5 週 | Discriminant analysis Fisher’s discriminant criterion Logistic regression KNN for classification |
| 第 6 週 | Maximum margin classifiers Kernel functions Support vector machines |
| 第 7 週 | Inter-school activity week (no class) |
| 第 8 週 | Model complexity and selection |
| 第 9 週 | Gaussian mixture models Density estimation Clustering: K-means and k-medoids |
| 第 10 週 | Hierarchical cluster analysis Number of classes determination Mode hunting |
| 第 11 週 | Principal component analysis Factor analysis Multidimensional scaling Locally linear embedding Independent component analysis |
| 第 12 週 | Linear time series analysis: Autocorrelation, Power spectrum White noise, stationarity, and ergodicity Multivariate series Linear Models |
| 第 13 週 | Multivariate AR model Statistical inference of model parameters Count and point process Implementation of multivariate AR model Granger causality AR and CCA calculations for Granger causality |
| 第 14 週 | Linear series with latent variables: State-space models Gaussian-process factor analysis Count and point series Bootstrapping for time series |
| 第 15 週 | Nonlinear concepts in time series analysis Detecting nonlinearity Nonparametric modeling Change point analysis Hidden Markov model |
| 第 16 週 | Nonlinear dynamical systems Map dynamics Recurrent neural networks Differential equations Attractors & chaos |
Advanced Data Analysis in Neuroscience: Integrating Statistical and Computational Models, Durstewitz, 2017. (Main, available online) Analysis of Neural Data, Kass, 2014. (Optional)
- 地點
- Online with with google meet.
- 時間
- Fridays 10am~11am by appointment.
- 聯絡方式
- Make an appointment by email by the Thursday evening.