Gaussian processes for machine learning

What Hacker News says

1. zeec123Oct 2023
   You may be interested in the book Gaussian Processes for Machine Learning: http://gaussianprocess.org/gpml/ (freely available)
   thread
2. steppiJan 2023
   Yes, there is a BayesianGaussianMixture class in sklearn.mixture [0]. There’s also sklearn.gaussian_process [1] which offers Gaussian process classification and regression, Bayesian learning algorithms which can be thought of as analogs of the support vector machine. Rasmussen’s Gaussian Processes for Machine Learning (2006) is a great introduction despite being a little old. [2].

   [0] https://scikit-learn.org/stable/modules/mixture.html

   [1] https://scikit-learn.org/stable/modules/gaussian_process.htm...

   [2] http://gaussianprocess.org/gpml/chapters/

   Show morethread
3. bhlSep 2019
   Implementing Gaussian Processes in Python is a fun way to practice your programming and mathematical abilities.

   I was working on a course project that used GPs for regression, with the inputs taken from the output of a CNN. We were trying to predict crop yield from satellite imagery, and GPs helped to incorporate spatio-temporal locality: think of predictions for neighboring county, or for image from a previous year.

   One issue we ran into while using the model from the original authors [1] was a memory constraint from building on top of Google Colab instead of a private AWS instance, and using a larger dataset to predict upon.

   The Gaussian kernel matrix grows O(n^2) per data point, and we had up to 22,000 datapoints to use. To get around the memory constraint, we had to re-implement the GP from scratch - carefully avoiding memory spikes that came with computing the kernel and mean predictions.

   It was a two-fold process. For understanding the exact equations, I would highly recommend the resource mentioned in the article [2], Gaussian Processes for Machine Learning Chapter 2. For a lecture video, I would recommend [3] which is from a ML course at Cornell.

   As for implementing GPs efficiently, this is still a challenge: numpy has a bunch of matrix-vector functions that are helpful, but there isn't a way to exploit the symmetry of the kernel matrix and use half the require memory.

   [1] https://aaai.org/ocs/index.php/AAAI/AAAI17/paper/view/14435/...

   [2] http://www.gaussianprocess.org/gpml/chapters/RW.pdf

   [3] https://www.youtube.com/watch?v=R-NUdqxKjos

   Show morethread
4. 0xmohitAug 2016
   Have you seen Gaussian Processes for Machine Learning [0]?

   The entire text is freely available online at the mentioned URL.

   [0] http://www.gaussianprocess.org/gpml/

   thread
5. phoenJan 2014
   The LION Way: Machine Learning plus Intelligent Optimization http://www.e-booksdirectory.com/details.php?ebook=9575

   A Course in Machine Learning http://www.e-booksdirectory.com/details.php?ebook=9395

   A First Encounter with Machine Learning http://www.e-booksdirectory.com/details.php?ebook=8818

   Bayesian Reasoning and Machine Learning http://www.e-booksdirectory.com/details.php?ebook=5283

   Introduction to Machine Learning http://www.e-booksdirectory.com/details.php?ebook=4493

   The Elements of Statistical Learning: Data Mining, Inference, and Prediction http://www.e-booksdirectory.com/details.php?ebook=3267

   Reinforcement Learning by C. Weber, M. Elshaw, N. M. Mayer http://www.e-booksdirectory.com/details.php?ebook=3227

   Machine Learning by Abdelhamid Mellouk, Abdennacer Chebira http://www.e-booksdirectory.com/details.php?ebook=2852

   How Are We To Know? by Nils J. Nilsson http://www.e-booksdirectory.com/details.php?ebook=2710

   Reinforcement Learning: An Introduction http://www.e-booksdirectory.com/details.php?ebook=1825

   Gaussian Processes for Machine Learning http://www.e-booksdirectory.com/details.php?ebook=1774

   Machine Learning, Neural and Statistical Classification http://www.e-booksdirectory.com/details.php?ebook=1118

   Introduction To Machine Learning http://www.e-booksdirectory.com/details.php?ebook=1117

   Inductive Logic Programming: Techniques and Applications http://www.e-booksdirectory.com/details.php?ebook=1105

   Practical Artificial Intelligence Programming in Java http://www.e-booksdirectory.com/details.php?ebook=32

   Information Theory, Inference, and Learning Algorithms http://www.e-booksdirectory.com/details.php?ebook=21

   Show morethread
6. netcanOct 2012
   Obviously the price of textbooks isn't the ultimate bottleneck for learning stuff. But, there are important difference between free & cheap. Easier access has all sorts of catalyzing effects.

   For example, a free textbook can be referred to. "For a more detailed discussion of this read chapter 4 of Gaussian Processes for Machine Learning."

   thread