If you are in the Tokyo area, I hope you will consider taking my classes! If you are not at the University of Tokyo or if you are not in academia, you will not be able to receive any credit, but you are still welcome to audit the classes. In this case I would appreciate if you could email me in advance so that I know how many class participants to expect. (Note: It's not me in the picture.)
This year I teach three courses:
Data Science for Practical Economic Research - from April 2017
Deep Learning and Related Methods for Large Dataset Information Processing - from September 2017
Topics in Asset Pricing - from September 2017
The details are below. My other teaching includes International Finance and International Trade.
Data Science for Practical Economic Research
Schedule: Tuesdays 13:00-14:45 and Wednesdays 13:00-14:45First meeting: Wednesday, April 5, 2017
Location: Classroom 3 (not 203), Economics Research Building (same as "Faculty of Economics, University of Tokyo" on Google Maps)
Course calalog codes: 291324-02, 0704254, 5123038. (In the catalog, the course title may sometimes show 'Applied Econometrics', for technical reasons. Hopefully this will not confuse you.)
Course content: This course is designed to help students use their time efficiently when performing economic data analysis. Topics include: Data manipulation: dataset transformation, visualization, data cleaning, web data scraping, conversion of data for the purposes of econometric estimation. Supervised machine learning: under-fitting and over-fitting, regularization, cross-validation, data augmentation. Unsupervised machine learning: clustering, factor analysis, principal component analysis, independent component analysis. Semi-supervised learning. Distributed data representation: word embedding. Nonlinear dimensionality reduction. Computational graphs and functional programming. Practical aspects of high-performance computing: GPU computing, cloud computing, model parallelism, and data parallelism. The course will include a first introduction to Python, TensorFlow, R, Scala, and Mathematica. For specialized tasks other software will be introduced. The students are encouraged to bring to the class their own datasets, which could then be used for the purposes of instruction and practical demonstration.
Deep Learning and Related Methods for Large Dataset Information Processing
Schedule: Tuesdays 10:25-12:10 and Fridays 10:25-12:10
First meeting: Tuesday, September 26, 2017
Classroom 8s, 1st floor, Akamon General Research Building (same as the building of "Library of Economics, University of Tokyo" on Google Maps). Note that I may request a classroom change before the first meeting. In that case it will be posted here.
Deep learning in artificial neural networks is a collection of statistical methods that benefit from large datasets and parallel computing. Recently it led to remarkable progress in many domains of research. This course provides an introduction to the subject, including the latest research. The structure of the course is chosen with the aim to be useful to students with very different academic backgrounds.
Topics include: Optimization: backpropagation, stochastic gradient descent and its accelerated versions, second-order optimization methods. Supervised and semi-supervised machine learning: under-fitting and over-fitting, regularization, cross-validation, data augmentation. Neural network architecture: activation functions and their properties, layer patterns. Training neural networks: data preprocessing, weight initialization, gradient flow, batch normalization, regularization, practical aspects of GPU computing and distributed training. Hyper-parameter optimization, model ensembles, model compression. Transfer learning and fine-tuning. Spatial data modeling: convolutional networks and their recent versions, visualizing their internal data representations, susceptibility to adversarial examples. Sequence data modeling: recurrent networks, LSTMs, GRUs, and their convolutional alternatives, attention. Recursive data modeling: recursive neural networks. Natural language processing: word embedding and its visualization, neural machine translation, speech recognition and synthesis. Unsupervised machine learning: autoencoders, graphical models, adversarial networks and their Nash equilibria. Deep reinforcement learning. Use of neural networks for designing and training other neural networks: neural architecture search, meta-learning. Hybrid computing combining advantages of neural networks and conventional computers. Use of deep learning for causal inference and counterfactual predictions. Privacy and ethical issues related to artificial intelligence.
Selected applications: econometric estimation of causal effects, solutions to game-theoretic models, economic time-series modeling, sentiment analysis, patient health outcome prediction, low-cost disease diagnosis, overcoming sensory loss with deep-learning technologies. The course will include a first introduction to Python and to deep learning software libraries TensorFlow, Keras, and Sonnet (and to some extent PyTorch/Torch, Theano, and Caffe2/Caffe). The precise selection of topics for the course will be adjusted based on the students' interests.
Topics in Asset Pricing
Schedule: Mondays 10:25-12:10 and Wednesdays 13:00-14:45
First meeting: Monday, September 25, 2017
Course content: Asset pricing theory and empirics - introduction and selected topics. Topics include: choice under uncertainty, static portfolio choice, capital asset pricing model, arbitrage pricing theory, stochastic discount factor, stock return predictability, consumption-based asset pricing, bond pricing, inter-temporal asset pricing, risk-sharing, asset markets with asymmetric information, household finance and its behavioral aspects.