Machine Learning Videos

• Aggregation
  • Overview of ensemble learning (boosting, blending, before and after the fact)
• Bayesian Learning
  • Validity of the Bayesian approach (prior, posterior, unknown versus probabilistic)
• Bias-Variance Tradeoff
  • Basic derivation (overfit and underfit, approximation-generalization tradeoff)
  • Example (sinusoidal target function)
  • Noisy case (Bias-variance-noise decomposition)
• Bin Model
  • Hoeffding Inequality (law of large numbers, sample, PAC)
  • Relation to learning (from bin to hypothesis, training data)
  • Multiple bins (finite hypothesis set, learning: search for green sample)
  • Union Bound (uniform inequality, M factor)
• Data Snooping
  • Definition and analysis (data contamination, model selection)
• Error Measures
  • User-specified error function (pointwise error, CIA, supermarket)
• Gradient Descent
  • Basic method (Batch GD) (first-order optimization)
  • Discussion (initialization, termination, local minima, second-order methods)
  • Stochastic Gradient Descent (the algorithm, SGD in action)
  • Initialization - Neural Networks (random weights, perfect symmetry)
• Learning Curves
  • Definition and illustration (complex models versus simple models)
  • Linear Regression example (learning curves for noisy linear target)
• Learning Diagram
  • Components of learning (target function, hypothesis set, learning algorithm)
  • Input probability distribution (unknown distribution, bin, Hoeffding)
  • Error measure (role in learning algorithm)
  • Noisy targets (target distribution)
  • Where the VC analysis fits (affected blocks in learning diagram)
• Learning Paradigms
  • Types of learning (supervised, reinforcement, unsupervised, clustering)
  • Other paradigms (review, active learning, online learning)
• Linear Classification
  • The Perceptron (linearly separable data, PLA)
  • Pocket algorithm (non-separable data, comparison with PLA)
• Linear Regression
  • The algorithm (real-valued function, mean-squared error, pseudo-inverse)
  • Generalization behavior (learning curves for linear regression)
• Logistic Regression
  • The model (soft threshold, sigmoid, probability estimation)
  • Cross entropy error (maximum likelihood)
  • The algorithm (gradient descent)
• Netflix Competition
  • Movie rating (singular value decomposition, essence of machine learning)
  • Applying SGD (stochastic gradient descent, SVD factors)
• Neural Networks
  • Biological inspiration (limits of inspiration)
  • Multilayer perceptrons (the model and its power and limitations)
  • Neural Network model (feedforward layers, soft threshold)
  • Backpropagation algorithm (SGD, delta rule)
  • Hidden layers (interpretation)
  • Regularization (weight decay, weight elimination, early stopping)
• Nonlinear Transformation
  • Basic method (linearity in the parameters, Z space)
  • Illustration (non-separable data, quadratic transform)
  • Generalization behavior (VC dimension of a nonlinear transform)
• Occam's Razor
  • Definition and analysis (definition of complexity, why simpler is better)
• Overfitting
  • The phenomenon (fitting the noise)
  • A detailed experiment (Legendre polynomials, types of noise)
  • Deterministic noise (target complexity, stochastic noise)
• Radial Basis Functions
  • Basic RBF model (exact interpolation, nearest neighbor)
  • K Centers (Lloyd's algorithm, unsupervised learning, pseudo-inverse)
  • RBF network (neural networks, local versus global, EM algorithm)
  • Relation to other techniques (SVM kernel, regularization)
• Regularization
  • Introduction (putting the brakes, function approximation)
  • Formal derivation (Legendre polynomials, soft-order constraint, augmented error)
  • Weight decay (Tikhonov, smoothness, neural networks)
  • Augmented error (proxy for out-of-sample error, choosing a regularizer)
  • Regularization parameter (deterministic noise, stochastic noise)
• Sampling Bias
  • Definition and analysis (Truman versus Dewey, matching the distributions)
• Support Vector Machines
  • SVM basic model (hard margin, constrained optimization)
  • The solution (KKT conditions, Lagrange, dual problem, quadratic programming)
  • Soft margin (non-separable data, slack variables)
  • Nonlinear transform (Z space, support vector pre-images)
  • Kernel methods (generalized inner product, Mercer's condition, RBF kernel)
• Validation
  • Introduction (validation versus regularization, optimistic bias)
  • Model selection (data contamination, validation set versus test set)
  • Cross Validation (leave-one-out, 10-fold cross validation)
• VC Dimension
• Growth function (dichotomies, Hoeffding Inequality)
• Examples (growth function for simple hypothesis sets)
• Break points (polynomial growth functions)
• Bounding the growth function (mathematical induction, polynomial bound)
• Definition of VC Dimension (shattering, distribution-free, Vapnik-Chervonenkis)
• VC Dimension of Perceptrons (number of parameters, lower and upper bounds)
• Interpreting the VC Dimension (degrees of freedom, Number of examples)
• 
  
• SOURCE: http://work.caltech.edu/library/

Learn Algo trading and Quantitative finance

Here is the list of online courses for learning algorithmic trading and quantitative finance:

FREE COURSES

• Machine Learning for Trading | Georgia Institute of Technology | Udacity
  Machine Learning for Trading | Udacity
• Investments | Massachusetts Institute of Technology
  http://ocw.mit.edu/courses/sloan...
• Investment Analysis | Carnegie Mellon University | Academic Financial Trading Platform
  http://www.academictrader.org/pr...
• Analytics of Finance | Massachusetts Institute of Technology
  http://ocw.mit.edu/courses/sloan...
• Topics in Mathematics with Applications in Finance | MIT
  http://ocw.mit.edu/courses/mathe...
• Options, Futures, and Other Derivatives | University of Toronto | John Hull
  The Academic Financial Trading Platform
• Model a Quantitative Trading Strategy in R | QuantInsti | DataCamp
  Model a Quantitative Trading Strategy in R
• The R Trader Blog
  http://www.thertrader.com/

PAID COURSES

• Trading with Python | Course/Blog
  Trading With Python | Become a quant.
• Trading Algorithms | Indian School of Business | Coursera
  Trading Algorithms | Coursera
• Advanced Trading Algorithms | Indian School of Business | Coursera
  Advanced Trading Algorithms | Coursera
• Design your own trading strategy | Indian School of Business | Coursera
  Design your own trading strategy – Culminating Project | Coursera
• Computational Investing | Georgia Institute of Technology | Coursera
  https://www.coursera.org/learn/c...
• Electronic Trading in Financial Markets | New York Institute of Finance | Edx
  https://www.edx.org/course/elect...
• Financial Engineering and Risk Management, Part 1 | University of Chicago | Coursera
  https://www.coursera.org/learn/f...
• Financial Engineering and Risk Management, Part 2 | University of Chicago | Coursera
  https://www.coursera.org/learn/f...
• Experfy | Algorithmic Trading Strategies | Nick Firoozye
  Algorithmic Trading Strategies
• Backtesting Algorithmic Trading Strategies | Quantor | Ernest P Chan
  Backtesting Online Course
• Introduction to quantitative investment | Quantor | Haksun Li
  Introduction to Quantitative Investment

PROGRAMS AND CERTIFICATES

• QuantInsti
  QuantInsti - Learn Algorithmic Trading from Market Practitioners
• Certificate in Quantitative Finance (CQF)
  http://www.cqf.com/
• Algorithmic Trading Strategies| Stevens Institute of Technology
  Algorithmic Trading Strategies
• Master of Financial Engineering Program| Berkeley-Haas
  http://mfe.berkeley.edu/

Beginner's starter kit to learn machine learning?!

[Below listed resources are taken from Quora , reposting it here for reference]

1. Assess, refresh and learn math and stats. This is probably a hard one, because I didn’t really have an idea what to look for. Nowadays, I’d advise you to go through this list (40+ Python Statistics For Data Science Resources) and learn as much as you can by applying the stuff with Python. For math, I’d consider taking linear algebra on Khan Academy.
2. Don’t be scared of investing in “theory”. What I hear a lot is that people don’t take the effort to go through some more dry/theoretical material. But I think that this is extremely valuable in the long run. I went through Machine Learning textbook and at the same time, I watched the videos of Machine Learning - Stanford University | Coursera. I think these were very instructive and gave a good solid basis to start from.
3. Get hands-on. These materials that I have mentioned are great to build up the foundations, but you should also be able to apply the concepts that you have learned about. You might consider taking Intro to Machine Learning Course | Udacity, but also take a look at Supervised Learning with scikit-learn and Unsupervised Learning in Python.
4. Practice. But, even more important than getting hands-on and revising the material with Python, is practicing. This step was probably the hardest one for me. I didn’t find any materials that I liked and that were on my level, so I had to do most of the things myself. Recently, I wrote a Python machine learning tutorial for beginners: Python Machine Learning: Scikit-Learn Tutorial. I’d also recommend Kaggle Python Tutorial on Machine Learning.
5. Don’t be scared of projects. After these, consider getting started on some projects via Your Home for Data Science. Make sure to put your code on Github so that others can see your progress and discuss.
6. Don’t stop. There are always new materials coming out and if you’re following some machine learning podcasts, you’ll always be up to date with the latest news. I’d also consider getting into R to do machine learning. The following courses can help you to do this: Free Introduction to R Programming Online Course, OpenIntro, Data Analysis and Statistical Inference, Basic Statistics or Statistics with R Track | DataCamp for the basics and statistics with R and Introduction to Machine Learning - Online Course, R: Unsupervised Learning and Machine Learning in R for beginners for machine learning.
7. Make use of all the material that is out there. Besides podcasts, there is great documentation out there that will most definitely help you to get out there and conquer the data fear: scikit-learn: machine learning in Python, Keras Documentation, and cheat sheets such as Choosing the right estimator , Scikit-Learn Cheat Sheet: Python Machine Learning, Keras Cheat Sheet: Neural Networks in Python, or other materials such as A visual introduction to machine learning, Clustering with Scikit with GIFs.

Baby steps for Machine learning

1. You can’t be a master without basics. Also you need to have the core skills in those domains.
   • To use machine learning first know how to program. The Google Class for Python
   • Learn and Understand statistics, especially Bayesian probability, which is essential for many machine learning algorithms. Blog for free books on data science / statistics.
   • Also learn Scipy, numpy, pandas from youtube
   
   
2. After basics, it’s time to take some free courses from best universities
   • Stanford’s Machine Learning Course, this is the famous course by Andrew Ng
   • Harvard’s Data Science Course, this is End-to-end data science course. Get practice with the entire data science workflow from data collection to analysis
   • Read An Introduction to Statistical Learning and Elements of Statistical Learning
   
   
3. Practice the entire machine learning workflow: Data collection, cleaning, and preprocessing. Model building, tuning, and evaluation using real data sets. Pick 5-10 datasets from the UCI Machine Learning Repository. For example, you can pick 3 datasets each for regression, classification, and clustering.
4. Now go and participate in machine learning competitions.