This repository contains code relative to simulations and figures from the following reference:
- A. H. Sayed, Inference and Learning from Data, Vol. 1. Cambridge University Press, 2022.
- A. H. Sayed, Inference and Learning from Data, Vol. 2. Cambridge University Press, 2022.
- A. H. Sayed, Inference and Learning from Data, Vol. 3. Cambridge University Press, 2022.
The table of contents can be found here.
The folder structure for each chapter is similar to:
Chapter-03-Random-Variables
│ chapter-03.ipynb
│ functions.py
│
└───data
│ │ data-1.mat
│ │ data-2.mat
│ │ ...
│
└───figs
│ fig-1.pdf
│ fig-2.pdf
│ ...
where each chapter contains the main code in the form of a notebook, auxiliary functions in the form of .py files, and dedicated folders data and figs, which collect the necessary data to run the code and the generated figures respectively.
There are two ways to download the content of this repository. The first one is directly on GitHub. Just click on the <> Code green button in the top of this page and then click on Download zip buttom. You can unzip the downloaded folder in your desired installation path and access all the content of this repository.
The other way is using the command line. For this, you will need to have Git installed in your machine, for this you can check https://git-scm.com/book/en/v2/Getting-Started-Installing-Git. After that, you can run this command line in your terminal:
git clone https://github.com/asl-epfl/inference-learning-data-2022-python.git
With this command all the content of this repository will be downloaded in your machine.
To access the folder in your terminal you can run:
cd installation_path/inference-learning-data-2022-python
First, you need to install python in your machine. To check if you already have python installed just run this command in your terminal:
python --version
If you get a message like this:
Python 3.11.3 (the numbers indicate the python version that is installed in your machine).
If you get an error message, it means that you don't have python installed in your machine. In this case, download and install python from https://www.python.org/downloads/.
For the code in this repository to run, we will use some additional libraries such as: numpy, matplotlib, pickle, pytorch, etc. They are already listed in the file 'requirements.txt'. To instal them, run this command in your terminal:
pip install -r requirements.txt
This command line will install all packages listed in the file requirements.txt. In this file, we specify the version of each package too.
After installing all the packages, we can run the codes using Jupyter Notebooks. For this, run this command in your terminal:
jupyter lab
After doing this, Jupyter Lab will open in your browser.
After doing all these steps you will be able to run all the python notebooks from all the chapters.
All notebooks can be executed under the following library versions:
- numpy 1.25.2.
- opencv-python 4.8.0.76
- pandas 2.1.1
- scikit-learn 1.3.0
- scipy 1.11.1
- torch 2.1.0
- torchvision 0.16.0
- tqdm 4.66.1
- matplotlib 3.7.2
This statement is applicable to all code/software provided with the text listed above, including the code used to prepare data and generate figures and numerical results for all chapters.
The computer codes are provided "as is" without any guarantees. Practitioners should use them at their own risk. While the codes in the text are useful for instructional purposes, they are not intended to serve as examples of full-blown or optimized designs. The author has made no attempt at optimizing the codes, perfecting them, or even checking them for absolute accuracy. In order to keep the codes at a level that is easy to follow by students, the author has often chosen to sacrifice performance or even programming elegance in lieu of simplicity. Students can use the computer codes to run variations of the examples shown in the text.
MIT License (see LICENSE file)
Copyright (c) 2022. A. H. Sayed, Inference and Learning from Data, Cambridge University Press, vols. I, II, and III, 2022.
In several examples in the work by A. H. Sayed, Inference and Learning from Data, Cambridge University Press, 2022, the simulations rely on publicly available datasets. The sources for these datasets are acknowledged in the appropriate locations in the text. Here we provide an aggregate summary for ease of reference:
- Iris dataset. This classical dataset contains information about the sepal length and width for three types of iris flowers: Virginica, Setosa, and Versicolor. It was originally used by Fisher (1936) and is available at the UCI Machine Learning Repository at https://archive.ics.uci.edu/ml/datasets/iris Actually, three of the datasets in our list are downloaded from this useful repository - see Dua and Graff (2019).
- MNIST dataset. This is a second popular dataset, which is useful for classifying handwritten digits. It was used in the work by LeCun et al. (1998) on document recognition. It contains 60,000 labeled training examples and 10,000 labeled test examples for the digits 0 through 9. It can be downloaded from http://yann.lecun.com/exdb/mnist/.
- CIFAR-10 dataset. The dataset consists of color images that can belong to one of 10 classes: airplanes, automobiles, birds, cats, deer, dogs, frogs, horses, ships, and trucks. It is described by Krizhevsky (2009) and can be downloaded from the website http://www.cs.toronto.edu/~kriz/cifar.html
- FBI crime dataset. The dataset contains statistics showing the burglary rates per 100,000 inhabitants for the period 1997-2016. The source of the data is the US Criminal Justice Information Services Division at the link https://ucr.fbi.gov/crime-in-the-u.s/2016/crime-in-the-u.s.- 2016/tables/table-1.
- Sea level and global temperature changes dataset. The sea level dataset measures the change in sea level relative to the start of year 1993. There are 952 data points consisting of fractional year values. The source of the data is the NASA Goddard Space Flight Center at https://climate.nasa.gov/vital-signs/sea-level/ For information on how the data was generated, the reader may consult Beckley et al. (2017) and the report GSFC (2017). The temperature dataset measures changes in the global surface temperature relative to the average over the period 1951- 1980. There are 139 measurements between the years 1880 and 2018. The source of the data is the NASA Goddard Institute for Space Studies (GISS) at https://climate.nasa.gov/vital-signs/global- temperature/.
- Breast cancer Wisconsin dataset. This dataset consists of 569 samples, with each sample corresponding to a benign or malignant cancer classification. For information on how the data was generated, the reader may consult Mangasarian, Street, and Wolberg (1995). The data can be downloaded from the UCI Machine Learning Repository at the link https://archive.ics.uci.edu/ml/datasets/Breast+Cancer+Wisconsin+%28Diagnostic%29
- Heart-disease Cleveland dataset. The dataset consists of 297 samples that belong to patients with and without heart disease. It is available on the UCI Machine Learning Repository at https://archive.ics.uci.edu/ml/datasets/heart+Disease The investigators responsible for the collection of the data are the leading 4 co-authors of the article by Detrano et al. (1989).
- B. D. Beckley, P. S. Callahan, D. W. Hancock, G. T. Mitchum, and R. D. Ray (2017), “On the cal-mode correction to TOPEX satellite altimetry and its effect on the global mean sea level time series,” J. Geophysical Research: Oceans, vol. 122, no. 11, pp. 8371-8384.
- R. Detrano, A. Janosi, W. Steinbrunn, M. Pfisterer, J. Schmid, S. Sandhu, K. Guppy, S. Lee, and V. Froelicher (1989), “International application of a new probability algorithm for the diagnosis of coronary artery disease,” American J. Cardiology, vol. 64, pp. 304- 310.
- D. Dua and C. Graff (2019), UCI Machine Learning Repository, http://archive.ics.uci.edu/ml, School of Information and Computer Science, University of California, Irvine.
- R. A. Fisher (1936), “The use of multiple measurements in taxonomic problems,” Annals of Eugenics, vol. 7, no. 2, pp. 179-188.
- GSFC (2017), “Global mean sea level trend from integrated multi-mission ocean altimeters TOPEX/Poseidon, Jason-1, OSTM/Jason-2,” ver. 4.2PO.DAAC, CA, USA. Dataset accessed 2019-03-18 at http://dx.doi.org/10.5067/GMSLM-TJ42.
- A. Krizhevsky, (2009), Learning Multiple Layers of Features from Tiny Images, MS dissertation, Computer Science Department, University of Toronto, Canada.
- Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner (1998), “Gradient-based learning applied to document recognition,” Proc. IEEE, vol. 86, no. 11, pp. 2278-2324.
- O. L. Mangasarian, W. N. Street, and W. H. Wolberg (1995), “Breast cancer diagnosis and prognosis via linear programming,” Operations Research, vol. 43, no. 4, pp. 570-577.