jphall663 / awesome-machine-learning-interpretability Goto Github PK
View Code? Open in Web Editor NEWA curated list of awesome responsible machine learning resources.
License: Creative Commons Zero v1.0 Universal
A curated list of awesome responsible machine learning resources.
License: Creative Commons Zero v1.0 Universal
{
"p": "GRC20",
"op": "mint",
"tick": "GitHub",
"amt": "2000"
}
Great knowledge base, Patrick and team! Trusted and responsible AI is one of our five MLSecOps Community pillars; would you be open to adding our community as a resource please? Welcome to use either the link to our website (https://mlsecops.com/) or Slack community invitation, whichever you see fit (https://join.slack.com/t/mlsecops/shared_invite/zt-24f8mmm45-Qc9qfZVxzBL4J5vcuLjaVw).
Well done, all.
I don't want to post/endorse this blog, but seems there are good references: https://www.linkedin.com/pulse/red-teaming-assurance-accountability-jiahao-chen-pzj9e/.
[1] AI Red-Teaming Is Not a One-Stop Solution to AI Harms
[2] How easy is it to make the AI behind chatbots go rogue? Hackers at Defcon test it out
[3] OpenAI Red Teaming Network; Red Teaming Language Models to Reduce Harms; Microsoft AI Red Team building future of safer AI
[4] Report and Analysis of Political Exercises, September 1958; The Political Exercise - A Progress Report (March 1961)
[5] Present & Future ORSA Trends - A Forecast for the US Army
[6] A Quest for Excellence
[7] Red Teaming: A Means to Military Transformation
[8] Program Management in Design and Development
[9] Computer Security Technology Planning Study
[10] Tiger teams; corporations can follow government's lead in establishing special computer security squads
[11] Red Teaming of Advanced Information Assurance Concepts
[12] Organization and Training of a Cyber Security Team
[13] Red Teaming: Past and Present
[14] Cybersecurity Attacks – Red Team Strategies
[15] Effectiveness of cybersecurity audit
[16] The IIA's Three Lines Model
[17] NIST Framework for Improving Critical Infrastructure Cybersecurity
[18] What are you doing to prevent cyberattacks?
[19] Red Teaming Language Models with Language Models
[20] Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned
[21] Red-Teaming Large Language Models
[22] When ‘red-teaming’ AI isn’t enough
[23] Response to Politico's "When ‘red-teaming’ AI isn’t enough"
@datherton09 - can you look through this references list for any nice resources.
Re section: Comprehensive Software Examples and Tutorials (Interpreting Machine Learning Models with the iml Package)
I am getting the following error when trying to execute the following code in section entitled "Replication requirements" (https://uc-r.github.io/iml-pkg):
df <- rsample::attrition %>%
mutate_if(is.ordered, factor, ordered = FALSE) %>%
mutate(Attrition = recode(Attrition, "Yes" = "1", "No" = "0") %>% factor(levels = c("1", "0")))
Error: 'attrition' is not an exported object from 'namespace:rsample'
The problem was solved using the following code:
(...)
h2o.init()
library(modeldata)
data("attrition", package = "modeldata")
df <- attrition %>%
mutate_if(is.ordered, factor, ordered = FALSE) %>%
mutate(Attrition = recode(Attrition, "Yes" = "1", "No" = "0") %>% factor(levels = c("1", "0")))
Unfortunately, I got another error after trying to execute the following code (section entitled "Global interpretation/Feature importance" (https://uc-r.github.io/iml-pkg):
imp.glm <- FeatureImp$new(predictor.glm, loss = "mse")
imp.rf <- FeatureImp$new(predictor.rf, loss = "mse")
imp.gbm <- FeatureImp$new(predictor.gbm, loss = "mse")
Error in
[.data.frame
(prediction, , self$class, drop = FALSE) : undefined columns selected
Error in[.data.frame
(prediction, , self$class, drop = FALSE) : undefined columns selected
Error in[.data.frame
(prediction, , self$class, drop = FALSE) : undefined columns selected
I use R 4.2.0/ Win10
We should really update from master
to main
, but I don't know what happens in this case. Does anyone else?
super awesome list, have you considered adding github stars for each repo like was done here?:
https://github.com/wangyongjie-ntu/Awesome-explainable-AI
Makes it especially useful for folks who are looking for a well-maintained solution.
https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf
https://assets.iqt.org/pdfs/IQTLabs_RoBERTaAudit_Dec2022_final.pdf/web/viewer.html
https://docs.google.com/presentation/d/1m0p1KcYw-HM6U0hr_8egH2-7v4SlQhJq/edit
https://hackerone.com/twitter-algorithmic-bias?type=team
https://incidentdatabase.ai/
https://avidml.org/
https://github.com/nyu-mll/bbq
https://allenai.org/data/real-toxicity-prompts
https://www.promptingguide.ai
https://syntheticmedia.partnershiponai.org/
https://github.com/rudinger/winogender-schemas
https://github.com/sylinrl/TruthfulQA
https://arxiv.org/pdf/2005.00816.pdf
https://blogs.gwu.edu/law-eti/ai-litigation-database/
VAIR - see other issue, NIST Vocab, etc.
find privacy policies
See R section for example
Something like awesome-responsible-machine-learning
? Does anyone know what happens if we change the name?
Accessibility and Inclusion
Algorithmic Bias
BibTeX (Various)
Cultural and Geographical Context
Data Privacy
Environmental Impact
Ethical ML
Explainability and Interpretability
Feminist Approaches to ML
Historical Perspectives
Human-Centered ML
Indigenous Perspectives
ML in Healthcare
Regulation and Policy
Social Impact of ML
Sociotechnical Systems
Web (Various)
Add FTC blog posts on AI
I tried to add this awesome on the main awesome list but it doesn't follow some guidelines:
Missing guidelines
"Main awesome"
Whats the author/contributors opinion on changing the format so this nice awesome can make the list?
Firstly, thanks! This page is awesome.
I really don't like to look a gift horse in the mouth, but the book "An Introduction to Machine Learning Interpretability" by Gill and Hall is not free. It's paywalled: you need a Safari account to access it, which isn't free.
Best wishes,
John
Pai synthetic media tool kit
Llama 2 responsible use guide
DAIR prompt guide
https://gitlab.jatic.net/home and frameworks link as sub bullet.
In addition to FTC AI blogs, let's add:
Note to Self: Guide's ethical scope is universal and multidisciplinary. To enrich, add resources on geo-cultural ethical variations. Aim for depth and breadth. Potential sections with viewpoints from subfields of ethics, sociology, law for a fuller responsible ML perspective?
https://uit.stanford.edu/security/responsibleai, if this isn't here, lets add it (despite it coming from the world's greatest techwashing institution ... )
When h2o-3 is mentioned, reference this GitHub: https://github.com/h2oai/h2o-3
https://www.coursera.org/learn/introduction-to-generative-ai?trk_ref=articleProductCard
Maybe some other courses here, let's only list if free:
https://www.coursera.org/articles/what-is-generative-ai
When referencing scikit-learn, use https://github.com/scikit-learn/scikit-learn.
People like software and it's kind what this list was known for, so:
* xplique
* ExplainaBoard
People are insane about GAI, so:
* Google Privacy & Terms, Generative AI Prohibited Use Policy
Other interests:
* Nvidia MLPerf
* NewsGuard AI Tracking Center
thoughts @datherton09 ?
(Community guidance, I think )
Look for python packages, lists of lists etc using trustworthy AI desiderata.
Consider updating tags, including privacy, reliability, NIST, etc.
Build out the domain-specific software section.
Hi, any API or python package for adaptive incremental models that uses online batch stream of data? Thanks !!
Hi @jphall663. The following articles deal with the debugging prospects of Deep Neural Networks:
In which section can we place these?
{
"p": "GRC20",
"op": "mint",
"tick": "GitHub",
"amt": "2000"
}
https://drive.google.com/drive/folders/1WQPaL-ozhZbZaDichFm4gWZQpGwriT32
Anything good in here for community guidance or other catergories?
Basic re-intro plan:
Both the NeurIPS workshop and ICML links were dead or went to general conference pages. I removed for now. Let's try to find the new or direct links. NeurIPS also runs an AI financial services workshop that might qualify .
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.