Machine learning is about constructing algorithms that can approximate complex functions from observations of input/output pairs. Machine learning is related to Statistics since its goal is to make predictions based on data.
Regression
The goal is to approximate a target function \(f\) or signal \(S\). The output space is often continuous.
Classification
The goal is to approximate a target function that assigns label to input points.
Applications of machine learning
Here are a few applications of machine learning. There are many more:
- Image classification
- Time-series prediction
- Language modeling
- Speech recognition
- Text translation
- Speech translation
- Content spam detection and filtering
- Fraud detection
Machine learning is very good at finding a function between two sets of data, no matter if such a function is grounded in reality or not. This is one of the many sources of algorithmic bias that can exist in ML models, which we need to understand and avoid when using such models in real-world settings.
Ethically debatable applications
- Patronym-based ethnicity classification (with the caveats it implies) (Bessudnov et al. 2021)
- Predicting gender, sexual orientation, etc. from biological data (Thomas et al. 2007; Leuner 2019; Clemens et al. 2022)
Bibliography
- Alexey Bessudnov, Denis Tarasov, Viacheslav Panasovets, Veronica Kostenko, Ivan Smirnov, Vladimir Uspenskiy. . "Predicting Ethnicity with Data on Personal Names in Russia". SocArXiv. DOI.
- Vince Thomas, Nitesh V. Chawla, Kevin W. Bowyer, Patrick J. Flynn. . "Learning to Predict Gender from Iris Images". In 2007 First IEEE International Conference on Biometrics: Theory, Applications, and Systems, 1–5. Crystal City, VA, USA: IEEE. DOI.
- John Leuner. . "A Replication Study: Machine Learning Models Are Capable of Predicting Sexual Orientation from Facial Images". arXiv. DOI.
- Benjamin Clemens, Jeremy Lefort-Besnard, Christoph Ritter, Elke Smith, Mikhail Votinov, Birgit Derntl, Ute Habel, Danilo Bzdok. . "Accurate Machine Learning Prediction of Sexual Orientation Based on Brain Morphology and Intrinsic Functional Connectivity". Cerebral Cortex. DOI.