Published on Sat Jun 06 2020

Detecting Emergent Intersectional Biases: Contextualized Word Embeddings Contain a Distribution of Human-like Biases

Wei Guo, Aylin Caliskan

State-of-the-art neural language modelsgenerate dynamic word embeddings. Current methods measure pre-defined social and intersectional biases. Bias at the intersection of race and gender associated with members of multiple minority groups have the highest magnitude.

2
8
19
Abstract

With the starting point that implicit human biases are reflected in the statistical regularities of language, it is possible to measure biases in English static word embeddings. State-of-the-art neural language models generate dynamic word embeddings dependent on the context in which the word appears. Current methods measure pre-defined social and intersectional biases that appear in particular contexts defined by sentence templates. Dispensing with templates, we introduce the Contextualized Embedding Association Test (CEAT), that can summarize the magnitude of overall bias in neural language models by incorporating a random-effects model. Experiments on social and intersectional biases show that CEAT finds evidence of all tested biases and provides comprehensive information on the variance of effect magnitudes of the same bias in different contexts. All the models trained on English corpora that we study contain biased representations. Furthermore, we develop two methods, Intersectional Bias Detection (IBD) and Emergent Intersectional Bias Detection (EIBD), to automatically identify the intersectional biases and emergent intersectional biases from static word embeddings in addition to measuring them in contextualized word embeddings. We present the first algorithmic bias detection findings on how intersectional group members are strongly associated with unique emergent biases that do not overlap with the biases of their constituent minority identities. IBD and EIBD achieve high accuracy when detecting the intersectional and emergent biases of African American females and Mexican American females. Our results indicate that biases at the intersection of race and gender associated with members of multiple minority groups, such as African American females and Mexican American females, have the highest magnitude across all neural language models.

Mon Nov 04 2019
Artificial Intelligence
Assessing Social and Intersectional Biases in Contextualized Word Representations
0
0
0
Fri Mar 05 2021
Artificial Intelligence
WordBias: An Interactive Visual Tool for Discovering Intersectional Biases Encoded in Word Embeddings
WordBias is an interactive visual tool designed to explore biases against intersectional groups encoded in static word embeddings. Given a pretrained static word embedding, WordBias computes the association of each word along different groups based on race, age, etc.
4
0
1
Thu Mar 11 2021
NLP
DebIE: A Platform for Implicit and Explicit Debiasing of Word Embedding Spaces
0
0
0
Fri Apr 26 2019
NLP
Are We Consistently Biased? Multidimensional Analysis of Biases in Distributional Word Vectors
Word embeddings have been shown to reflect many of the pronounced societal biases (e.g., gender bias or racial bias) Existing studies are limited in scope and do not investigate the consistency of biasesacross relevant dimensions like embedding models, types of texts, and different languages.
0
0
0
Thu Apr 18 2019
NLP
Analytical Methods for Interpretable Ultradense Word Embeddings
Word embeddings are useful for a wide variety of tasks, but they lack horriblyinterpretable dimensions. By rotating word spaces, interpretable dimensions can be identified while preserving the information contained in the embedDings. DensRay is a new method we propose.
0
0
0
Thu Sep 06 2018
Artificial Intelligence
Uncovering divergent linguistic information in word embeddings with lessons for intrinsic and extrinsic evaluation
A new perspective on how embeddings encode divergent linguistic information. Effect of our transformations in downstream tasks is higher for unsupervised systems than for supervised ones.
0
0
0
Tue Aug 28 2018
NLP
Understanding Back-Translation at Scale
Back-translations obtained via sampling or noised beam outputs are most effective. We scale to hundreds of millions of monolingual sentences and achieve a new state of the art of 35 BLEU.
1
587
1,633
Mon Jun 12 2017
NLP
Attention Is All You Need
The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms. Experiments on two machine translation tasks show these models to be superior in
50
215
883
Thu May 28 2020
NLP
Language Models are Few-Shot Learners
GPT-3 is an autoregressive language model with 175 billion parameters, 10x more than any previous non-sparse language model. It can perform tasks without gradient updates or fine-tuning, with tasks and few-shot demonstrations specified purely via text interaction with the model.
20
21
235
Wed Jun 05 2019
NLP
Energy and Policy Considerations for Deep Learning in NLP
Recent progress in hardware and methodology for training neural networks has ushered in a new generation of large networks trained on abundant data. These models have obtained notable gains in accuracy across many NLP tasks. As a result these models are costly to train and develop.
19
16
79
Thu Oct 11 2018
NLP
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
BERT is designed to pre-train deep                bidirectional representations from unlabeled text. It can be fine-tuned with just one additional output layer to create state-of-the-art models for a wide range of tasks.
13
8
15
Wed Jun 19 2019
Machine Learning
XLNet: Generalized Autoregressive Pretraining for Language Understanding
XLNet is a generalized autoregressive pretraining method. It enables learning bidirectional contexts by maximizing the expected likelihood over all permutations of the factorization order. Under comparable experiment settings, XLNet outperforms BERT on 20 tasks, often by a large margin.
1
8
9
Thu Jun 24 2021
NLP
Towards Understanding and Mitigating Social Biases in Language Models
As machine learning methods are deployed in real-world settings such as healthcare, legal systems, and social science, it is crucial to recognize how they shape social biases and stereotypes. Large-scale pretrained language models (LMs) can be potentially dangerous in manifesting harmful representational biases.
11
50
168
Fri Mar 05 2021
Artificial Intelligence
WordBias: An Interactive Visual Tool for Discovering Intersectional Biases Encoded in Word Embeddings
WordBias is an interactive visual tool designed to explore biases against intersectional groups encoded in static word embeddings. Given a pretrained static word embedding, WordBias computes the association of each word along different groups based on race, age, etc.
4
0
1
Wed Oct 28 2020
Computer Vision
Image Representations Learned With Unsupervised Pre-Training Contain Human-like Biases
Machine learning models can automatically learn bias from the way people are stereotypically portrayed on the web. For the first time in unsupervised computer vision, we quantify implicit human biases about weight, disabilities, and several ethnicities.
1
0
0
Tue Apr 13 2021
Artificial Intelligence
Semantic maps and metrics for science Semantic maps and metrics for science using deep transformer encoders
0
0
0
Tue Nov 24 2020
NLP
Unequal Representations: Analyzing Intersectional Biases in Word Embeddings Using Representational Similarity Analysis
Black women are simultaneously less feminine than White women, and less Black than Black men. This finding aligns with intersectionality theory, which argues that multiple identity categories (such as race or sex) layer on top of each other.
0
0
0