![fly51fly's tweet photo. [CL] Bias and Fairness in Large Language Models: A Survey
I O. Gallegos, R A. Rossi, J Barrow, M M Tanjim… [Stanford University & Adobe Research] (2023)
https://t.co/UVkmY3U1TB
- Provides definitions and a taxonomy of social biases and harms relevant to NLP, including representational and allocational harms.
- Proposes a taxonomy of metrics to evaluate bias using embeddings, probabilities, or generated text. Discusses limitations of embedding metrics.
- Categorizes datasets by counterfactual inputs or prompts. Highlights reliability and validity issues in existing datasets. Releases public datasets.
- Classifies techniques to mitigate bias via pre-processing, in-training, intra-processing, or post-processing interventions.
- Identifies challenges including addressing power imbalances, conceptualizing fairness, improving evaluation principles, expanding mitigation efforts, and exploring theoretical limits.
- Calls for centering marginalized communities, participatory research, reconstructing assumptions, expanding resources, developing fairness desiderata, rethinking social groups, examining reliability and validity, considering benefits and harms of benchmarks, expanding evaluation, enabling scalability of mitigations, developing hybrid techniques, understanding bias encoding in models, and establishing guarantees.](https://pbs.twimg.com/media/F5UqKDwW8AAnaAj.jpg)
![fly51fly's tweet photo. [CL] Bias and Fairness in Large Language Models: A Survey
I O. Gallegos, R A. Rossi, J Barrow, M M Tanjim… [Stanford University & Adobe Research] (2023)
https://t.co/UVkmY3U1TB
- Provides definitions and a taxonomy of social biases and harms relevant to NLP, including representational and allocational harms.
- Proposes a taxonomy of metrics to evaluate bias using embeddings, probabilities, or generated text. Discusses limitations of embedding metrics.
- Categorizes datasets by counterfactual inputs or prompts. Highlights reliability and validity issues in existing datasets. Releases public datasets.
- Classifies techniques to mitigate bias via pre-processing, in-training, intra-processing, or post-processing interventions.
- Identifies challenges including addressing power imbalances, conceptualizing fairness, improving evaluation principles, expanding mitigation efforts, and exploring theoretical limits.
- Calls for centering marginalized communities, participatory research, reconstructing assumptions, expanding resources, developing fairness desiderata, rethinking social groups, examining reliability and validity, considering benefits and harms of benchmarks, expanding evaluation, enabling scalability of mitigations, developing hybrid techniques, understanding bias encoding in models, and establishing guarantees.](https://pbs.twimg.com/media/F5UqKDwWEAA2UKW.jpg)
![fly51fly's tweet photo. [CL] Bias and Fairness in Large Language Models: A Survey
I O. Gallegos, R A. Rossi, J Barrow, M M Tanjim… [Stanford University & Adobe Research] (2023)
https://t.co/UVkmY3U1TB
- Provides definitions and a taxonomy of social biases and harms relevant to NLP, including representational and allocational harms.
- Proposes a taxonomy of metrics to evaluate bias using embeddings, probabilities, or generated text. Discusses limitations of embedding metrics.
- Categorizes datasets by counterfactual inputs or prompts. Highlights reliability and validity issues in existing datasets. Releases public datasets.
- Classifies techniques to mitigate bias via pre-processing, in-training, intra-processing, or post-processing interventions.
- Identifies challenges including addressing power imbalances, conceptualizing fairness, improving evaluation principles, expanding mitigation efforts, and exploring theoretical limits.
- Calls for centering marginalized communities, participatory research, reconstructing assumptions, expanding resources, developing fairness desiderata, rethinking social groups, examining reliability and validity, considering benefits and harms of benchmarks, expanding evaluation, enabling scalability of mitigations, developing hybrid techniques, understanding bias encoding in models, and establishing guarantees.](https://pbs.twimg.com/media/F5UqKDvXkAAuZXK.jpg)
Olivia
Online
Network Repository
@networkrepo
A data repository with interactive visual analytics for real-time exploration. #dataviz #networkscience #machinelearning #graphs #networks #bigdata #socialmedia
San Francisco, CA
Joined April 2014
1.5K Following
473 Followers
88 Posts
Pinned Tweet
Large Generative Graph Models (LGGMs): A Text to Graph Generative Model Trained on Graphs from Many Different Domains (https://t.co/DA52QFAsN4)
PDFTriage: Question Answering over Long, Structured Documents
paper page: https://t.co/UvaJIXXCyH
Large Language Models (LLMs) have issues with document question answering (QA) in situations where the document is unable to fit in the small context length of an LLM. To overcome this issue, most existing works focus on retrieving the relevant context from the document, representing them as plain text. However, documents such as PDFs, web pages, and presentations are naturally structured with different pages, tables, sections, and so on. Representing such structured documents as plain text is incongruous with the user's mental model of these documents with rich structure. When a system has to query the document for context, this incongruity is brought to the fore, and seemingly trivial questions can trip up the QA system. To bridge this fundamental gap in handling structured documents, we propose an approach called PDFTriage that enables models to retrieve the context based on either structure or content. Our experiments demonstrate the effectiveness of the proposed PDFTriage-augmented models across several classes of questions where existing retrieval-augmented LLMs fail. To facilitate further research on this fundamental problem, we release our benchmark dataset consisting of 900+ human-generated questions over 80 structured documents from 10 different categories of question types for document QA.
Who to follow
Jean-Gabriel Young 
@_jgyou
AI @ https://t.co/S4ft6LXcdt /
Assistant Prof. Complex Systems @ UVM (on leave) /
Lab: https://t.co/GSJzkXRaE3 /
Zachary's Karate Club Club Member Est. 2020 🏆
Ronaldo Menezes
@ronaldomenezes
Prof of Data and Network Science at @UniofExeter, @exetercompsci. Editor-in-Chief of @appliednetsci.
Adilson E. Motter
@adilson_motter
Complex systems and networks. Professor of Physics and Director of the Center for Network Dynamics, Northwestern University: https://t.co/0JCwDsoA4l
[CL] Bias and Fairness in Large Language Models: A Survey
I O. Gallegos, R A. Rossi, J Barrow, M M Tanjim… [Stanford University & Adobe Research] (2023)
https://t.co/UVkmY3U1TB
- Provides definitions and a taxonomy of social biases and harms relevant to NLP, including representational and allocational harms.
- Proposes a taxonomy of metrics to evaluate bias using embeddings, probabilities, or generated text. Discusses limitations of embedding metrics.
- Categorizes datasets by counterfactual inputs or prompts. Highlights reliability and validity issues in existing datasets. Releases public datasets.
- Classifies techniques to mitigate bias via pre-processing, in-training, intra-processing, or post-processing interventions.
- Identifies challenges including addressing power imbalances, conceptualizing fairness, improving evaluation principles, expanding mitigation efforts, and exploring theoretical limits.
- Calls for centering marginalized communities, participatory research, reconstructing assumptions, expanding resources, developing fairness desiderata, rethinking social groups, examining reliability and validity, considering benefits and harms of benchmarks, expanding evaluation, enabling scalability of mitigations, developing hybrid techniques, understanding bias encoding in models, and establishing guarantees.
![fly51fly's tweet photo. [CL] Bias and Fairness in Large Language Models: A Survey
I O. Gallegos, R A. Rossi, J Barrow, M M Tanjim… [Stanford University & Adobe Research] (2023)
https://t.co/UVkmY3U1TB
- Provides definitions and a taxonomy of social biases and harms relevant to NLP, including representational and allocational harms.
- Proposes a taxonomy of metrics to evaluate bias using embeddings, probabilities, or generated text. Discusses limitations of embedding metrics.
- Categorizes datasets by counterfactual inputs or prompts. Highlights reliability and validity issues in existing datasets. Releases public datasets.
- Classifies techniques to mitigate bias via pre-processing, in-training, intra-processing, or post-processing interventions.
- Identifies challenges including addressing power imbalances, conceptualizing fairness, improving evaluation principles, expanding mitigation efforts, and exploring theoretical limits.
- Calls for centering marginalized communities, participatory research, reconstructing assumptions, expanding resources, developing fairness desiderata, rethinking social groups, examining reliability and validity, considering benefits and harms of benchmarks, expanding evaluation, enabling scalability of mitigations, developing hybrid techniques, understanding bias encoding in models, and establishing guarantees.](https://pbs.twimg.com/media/F5UqKDyXoAAf1KX.jpg)
Call for participation
IPDPS GrAPL 2021: IPDPS Workshop on Graphs, Architectures, Programming, and Learning - May 17th, 2021
#graphs #GNN #graphdatabase #MachineLearning
https://t.co/nt0LBSjPWL
Just came across the amazing @networkrepo https://t.co/1ccNMEl0xw: A huge network data repository with interactive visualization and mining. Thank you for the effort @n_kahmed and Ryan Rossi!
From Community to Role-based Graph Embeddings
“differences between roles and communities, and formalize the general mechanisms (e.g., random walks, feature diffusion) that give rise to community or role-based embeddings”
https://t.co/U1j0Mg3UYn
Looking for dynamic/temporal network datasets? Check out Network Repository! https://t.co/aZxZ52Jhff #networkscience #graphs #datascience #machinelearning #neuroscience #HPC
Check out Graphvis! our new graph visualization and analytics tool from @networkrepo https://t.co/hr0cCuip9F #networkscience #graphs #datascience #machinelearning #neuroscience #networks #networkrepository
Check the new brain network collection. #networkscience #graphs #datascience #machinelearning #neuroscience https://t.co/aMLlobhbiD
Learning Role-based Graph Embeddings. https://t.co/D0F9BIqNJ9
Our large-scale network classification analysis with thousands of networks is accepted @CompleNet 2018 #CompleNet18 #networkscience #networks #MachineLearning https://t.co/kAjubCV26i
Network Classification and Categorization using Network Repository https://t.co/aeSyRm88cm #networkscience #MachineLearning #ComplexNetworks https://t.co/hOB0sMDSSk
Network Classification and Categorization using Network Repository https://t.co/aeSyRm88cm #networkscience #MachineLearning #ComplexNetworks https://t.co/hOB0sMDSSk
Network Classification and Categorization. (arXiv:1709.04481v1 [cs.SI]) https://t.co/EsOdcTGAba
We envision interactive data repositories,to make data accessible with tools for interactive data mining #openaccess https://t.co/gsW3hpSq0p
"On Sampling from Massive Graph Streams. (arXiv:1703.02625v1 [cs.SI])" https://t.co/MDJNuEPVu4
Trends for you
Most Popular Users
Elon Musk
@elonmusk
241.2M followers
Barack Obama
@barackobama
119.1M followers
Cristiano Ronaldo
@cristiano
112.8M followers
Donald J. Trump
@realdonaldtrump
111.8M followers
Narendra Modi
@narendramodi
107.1M followers
Rihanna
@rihanna
98.3M followers
NASA
@nasa
92.3M followers
Justin Bieber
@justinbieber
91.4M followers
KATY PERRY
@katyperry
89M followers
Taylor Swift
@taylorswift13
82.9M followers
Lady Gaga
@ladygaga
74.4M followers
Virat Kohli
@imvkohli
71.8M followers
Kim Kardashian
@kimkardashian
70.4M followers
YouTube
@youtube
68.8M followers
Neymar Jr
@neymarjr
64.8M followers
Bill Gates
@billgates
64.6M followers
The Ellen Show
@theellenshow
62.4M followers
Selena Gomez
@selenagomez
62M followers
CNN
@cnn
61.8M followers
X
@x
60.8M followers