Fairness

Netflix Prize consists of about 100,000,000 ratings for 17,770 movies given by 480,189 users. Each rating in the training dataset consists of four entries: user, movie, date of grade, grade. Users and movies are represented with integer IDs, while ratings range from 1 to 5. Source: The Netflix Prize Image Source: https://www.netflixprize.com/

The UTKFace dataset is a large-scale face dataset with long age span (range from 0 to 116 years old). The dataset consists of over 20,000 face images with annotations of age, gender, and ethnicity. The images cover large variation in pose, facial expression, illumination, occlusion, resolution, etc. This dataset could be used on a variety of tasks, e.g., face detection, age estimation, age progression/regression, landmark localization, etc. Source: https://susanqq.github.io/UTKFace/ Image Source: https://susanqq.github.io/UTKFace/

FairFace is a face image dataset which is race balanced. It contains 108,501 images from 7 different race groups: White, Black, Indian, East Asian, Southeast Asian, Middle Eastern, and Latino. Images were collected from the YFCC-100M Flickr dataset and labeled with race, gender, and age groups. Source: https://github.com/joojs/fairface

MORPH is a facial age estimation dataset, which contains 55,134 facial images of 13,617 subjects ranging from 16 to 77 years old. Source: Deep Ordinal Regression Forests Image Source: https://uncw.edu/oic/tech/morph.html

WinoBias contains 3,160 sentences, split equally for development and test, created by researchers familiar with the project. Sentences were created to follow two prototypical templates but annotators were encouraged to come up with scenarios where entities could be interacting in plausible ways. Templates were selected to be challenging and designed to cover cases requiring semantics and syntax separately. Source: WinoBias Image Source: https://uclanlp.github.io/corefBias/overview

To validate the racial bias of four commercial APIs and four state-of-the-art (SOTA) algorithms. Source: Racial Faces in-the-Wild: Reducing Racial Bias by Information Maximization Adaptation Network

The General Video Game AI (GVGAI) framework is widely used in research which features a corpus of over 100 single-player games and 60 two-player games. These are fairly small games, each focusing on specific mechanics or skills the players should be able to demonstrate, including clones of classic arcade games such as Space Invaders, puzzle games like Sokoban, adventure games like Zelda or game-theory problems such as the Iterative Prisoners Dilemma. All games are real-time and require players to make decisions in only 40ms at every game tick, although not all games explicitly reward or require fast reactions; in fact, some of the best game-playing approaches add up the time in the beginning of the game to run Breadth-First Search in puzzle games in order to find an accurate solution. However, given the large variety of games (many of which are stochastic and difficult to predict accurately), scoring systems and termination conditions, all unknown to the players, highly-adaptive general methods are needed to tackle the diverse challenges proposed. Source: Rolling Horizon Evolutionary Algorithms for General Video Game Playing Image Source: http://www.gvgai.net/

The HELP dataset is an automatically created natural language inference (NLI) dataset that embodies the combination of lexical and logical inferences focusing on monotonicity (i.e., phrase replacement-based reasoning). The HELP (Ver.1.0) has 36K inference pairs consisting of upward monotone, downward monotone, non-monotone, conjunction, and disjunction. Source: HELP

MIAP is a dataset created by obtaining a new set of annotations on a subset of the Open Images dataset, containing bounding boxes and attributes for all of the people visible in those images, as the original Open Images dataset annotations are not exhaustive, with bounding boxes and attribute labels for only a subset of the classes in each image. The MIAP dataset focuses on enabling ML Fairness research. It provides additional annotations for 100,000 (70k from training and 30k from validation/test) images that contain at least one person bounding box in the original annotations. These additional annotations provide exhaustive bounding boxes for all people in an image. Person boxes are further annotated with attribute labels for fairness research. Annotated attributes include the human perceived gender presentation (predominantly feminine, predominantly masculine, and unknown) and perceived age range (young, middle, older, and unknown) of the localized person. This procedure adds nearly 100,000 new boxes that were not annotated under the original labeling pipeline. Annotations on the exhaustive set enable research into the fairness properties of models trained on partial annotations and the pipelines that produce these annotations.

ACS PUMS stands for American Community Survey (ACS) Public Use Microdata Sample (PUMS) and has been used to construct several tabular datasets for studying fairness in machine learning: - ACSIncome: to predict whether an individual’s income is above $50,000. - ACSPublicCoverage: to predict whether an individual is covered by public health insurance. - ACSMobility: to predict whether an individual had the same residential address one year ago. - ACSEmployment: to predict whether an individual is employed. - ACSTravelTime: predict whether an individual has a commute to work that is longer than 20 minutes.

A new face annotation dataset with balanced distribution between genders and ethnic origins. Source: SensitiveNets: Learning Agnostic Representations with Application to Face Images

Bank Account Fraud (BAF) is a large-scale, realistic suite of tabular datasets. The suite was generated by applying state-of-the-art tabular data generation techniques on an anonymized, real-world bank account opening fraud detection dataset. Source: Turning the Tables: Biased, Imbalanced, Dynamic Tabular Datasets for ML Evaluation

SPEECH-COCO contains speech captions that are generated using text-to-speech (TTS) synthesis resulting in 616,767 spoken captions (more than 600h) paired with images. Source: SPEECH-COCO: 600k Visually Grounded Spoken Captions Aligned to MSCOCO Data Set

CI-MNIST (Correlated and Imbalanced MNIST) is a variant of MNIST dataset with introduced different types of correlations between attributes, dataset features, and an artificial eligibility criterion. For an input image $x$, the label $y \in \\{1, 0\\}$ indicates eligibility or ineligibility, respectively, given that $x$ is even or odd. The dataset defines the background colors as the protected or sensitive attribute $s \in \\{0, 1\\}$, where blue denotes the unprivileged group and red denotes the privileged group. The dataset was designed in order to evaluate bias-mitigation approaches in challenging setups and be capable of controlling different dataset configurations.

This research aimed at the case of customers default payments in Taiwan and compares the predictive accuracy of probability of default among six data mining methods. From the perspective of risk management, the result of predictive accuracy of the estimated probability of default will be more valuable than the binary result of classification - credible or not credible clients. Because the real probability of default is unknown, this study presented the novel Sorting Smoothing Method to estimate the real probability of default. With the real probability of default as the response variable (Y), and the predictive probability of default as the independent variable (X), the simple linear regression result (Y = A + BX) shows that the forecasting model produced by artificial neural network has the highest coefficient of determination; its regression intercept (A) is close to zero, and regression coefficient (B) to one. Therefore, among the six data mining techniques, artificial neural network is the only one that can accurately estimate the real probability of default.

The Dialogue Fairness dataset is used to evaluate and understand fairness in dialogue models, focusing on gender and racial biases. Source: https://github.com/zgahhblhc/DialogueFairness

ImDrug is a comprehensive benchmark with an open-source Python library which consists of 4 imbalance settings, 11 AI-ready datasets, 54 learning tasks and 16 baseline algorithms tailored for imbalanced learning. It features modularized components including formulation of learning setting and tasks, dataset curation, standardized evaluation, and baseline algorithms. It also provides an accessible and customizable testbed for problems and solutions spanning a broad spectrum of the drug discovery pipeline such as molecular modeling, drug-target interaction and retrosynthesis.

KANFace consists of 40K still images and 44K sequences (14.5M video frames in total) captured in unconstrained, real-world conditions from 1,045 subjects. The dataset is manually annotated in terms of identity, exact age, gender and kinship. Source: KANFace Dataset

TwinViews-13k is a dataset of 13,855 pairs of left-leaning and right-leaning political statements, each pair matched by topic. It was created to study political bias in reward and language models, with a focus on understanding the interaction between model alignment to truthfulness and the emergence of political bias. The dataset was generated using GPT-3.5 Turbo, with extensive auditing to ensure ideological balance and topical relevance. This dataset can be used for various tasks related to political bias, natural language processing, and model alignment, particularly in studies examining how political orientation impacts model outputs.

ec-darkpattern is a dataset for dark pattern detection and prepared its baseline detection performance with state-of-the-art machine learning methods. The original dataset was obtained from Mathur et al.’s study in 2019 [11kScale], which consists of 1,818 dark pattern texts from shopping sites. Negative samples, i.e., non-dark pattern texts, by retrieving texts from the same websites as Mathur et al.'s dataset. Source: Dark patterns in e-commerce: a dataset and its baseline evaluations Image Source: https://arxiv.org/pdf/2211.06543v1.pdf

A Racial Fairness Benchmark Dataset for Face Forgery Detection.

A maintained database tracks ICLR submissions and reviews, augmented with author profiles and higher-level textual features.

The Tufts fNIRS to Mental Workload (fNIRS2MW) open-access dataset is a new dataset for building machine learning classifiers that can consume a short window (30 seconds) of multivariate fNIRS recordings and predict the mental workload intensity of the user during that window. You can use this dataset for tasks like - time series classification using sliding windows - domain adaptation or domain generalization (how well does your classifier generalize to a new subject?) - fairness of time series classifiers (does performance of your classifier vary by subject race or gender?) Useful Links: * Project Website (and data download links): https://tufts-hci-lab.github.io/code_and_datasets/fNIRS2MW.html * Code for benchmarks: https://github.com/tufts-ml/fNIRS-mental-workload-classifiers * DataSheet documentation: https://github.com/tufts-ml/fNIRS-mental-workload-classifiers/blob/main/Datasheet-Tufts-fNIRS2MW.pdf * Academic Paper (published at NeurIPS Datasets & Benchmarks '21) https://datasets-benchmarks-proceedings.neurips.cc/paper/2021/file/bd686fd640be98efaae0091fa301e613-Paper-round2.pdf ## Motivation We are interested in building brain computer interfaces (BCIs) that would help out everyday computer users working at a desktop or laptop. In our target future use case, a user would actively use a keyboard and mouse as usual, but also wear a non-intrusive headband sensor that would passively provide real-time measurements of brain activity to the computer. Based on moment-to-moment estimates of mental workload, the computer could adjust the interface to support the user. Functional near-infrared spectroscopy (fNIRS) is a promising sensor technology for achieving this goal of "everyday BCI", compared to alternatives like EEG or fMRI. We have developed a prototype fNIRS probe mounted on a headband that we used to collect this dataset (see our paper for details). # Dataset Overview For a complete dataset summary, see our public DataSheet PDF For each participant (68 recommended; 87 total), the dataset contains the following records obtained during one 30-60 minute experimental session. Each subject contributes just over 21 minutes of fNIRS data from the desired n-back experimental conditions, with remaining time related to rest or instruction periods. - fNIRS recordings - Multivariate (D=8) time-series representing brain activity throughout the session, recorded by a sensor probe placed on the forehead and secured via headband - All measurements are recorded at a regular sampling rate of 5.2 Hz. - At each timestep, we record 8 real-valued measurements, one for each combination of - 2 blood chemical concentration changes (oxygenated hemoglobin and deoxygenated hemoglobin) - 2 optical data types used for the measurement (intensity and phase) - 2 spatial locations on the forehead. - The units of each measurement are micro-moles of (oxy-/deoxy-)hemoglobin per liter of tissue. - Activity labels - Annotations of the experimental task activity the subject performed throughout the session, including instruction, rest, and active experiment segments. - We label each segment of the active experiment as one of four possible n-back working memory intensity levels (0-back, 1-back, 2-back, or 3-back). Increased intensity levels are intended to induce an increased level of cognitive workload. - For all experiments reported in the paper, we focus on a binary task (0 vs 2 back) - Demographics - The participant’s age, gender, race, handedness, and other attributes. This lets us measure and audit performance by subpopulation (e.g. how does the classifier perform on white subjects vs. black subjects). ## Publications The Tufts fNIRS Mental Workload Dataset & Benchmark for Brain-Computer Interfaces that Generalize Zhe Huang, Liang Wang, Giles Blaney, Christopher Slaughter, Devon McKeon, Ziyu Zhou, Robert Jacob, and Michael C. Hughes To appear in the Proceedings of Neural Information Processing Systems (NeurIPS 2021) Track on Datasets and Benchmarks , 2021 Link to Paper PDF: <a href="https://openreview.net/pdf?id=QzNHE7QHhut">https://openreview.net/pdf?id=QzNHE7QHhut</a>