Hacker News new | past | comments | ask | show | jobs | submit | _a0vv's comments login

Thanks, I modified the title to include the year. It's old but a nice homage to Terry Jones who passed away recently. Plus the article is still relevant today I think.


Abstract:

Black-box explanation is the problem of explaining how a machine learning model -- whose internal logic is hidden to the auditor and generally complex -- produces its outcomes. Current approaches for solving this problem include model explanation, outcome explanation as well as model inspection. While these techniques can be beneficial by providing interpretability, they can be used in a negative manner to perform fairwashing, which we define as promoting the false perception that a machine learning model respects some ethical values. In particular, we demonstrate that it is possible to systematically rationalize decisions taken by an unfair black-box model using the model explanation as well as the outcome explanation approaches with a given fairness metric. Our solution, LaundryML, is based on a regularized rule list enumeration algorithm whose objective is to search for fair rule lists approximating an unfair black-box model. We empirically evaluate our rationalization technique on black-box models trained on real-world datasets and show that one can obtain rule lists with high fidelity to the black-box model while being considerably less unfair at the same time.


The same could be said about archaeologists discovering a lost city, since it wasn't always "lost" and was certainly discovered many times before. I guess the timescale gives the impression of a "fake discovery".


The ban only concerns "students under the age of 15", who typically take notes on paper. The teacher can also make an exception for pedagogical purpose.

> There are exceptions in place for students with disabilities and for the educational use of devices in the classroom and in extra-curricular activities


Sure but they also use Google and Wikipedia etc.


Under 15? Not really, can use purely offline sources for that generic material. Good skill as well. Still should have a basic computing class though, perhaps wrapped into a library class.


In my experience all printed material is incomplete, obsolete, or one-sided and frequently all three.

That is not to talk about materials that can't be encoded as text, such as instructions and how-tos for whatever the French version of shop-class is.

You would propose that students be taught such subjects as the French revolution, based on what materials are available in the school library? How can you even ethically do that, when the kids probably can't even get the complete text of the Tennis court oath?


I very much doubt the historical consensus about French revolution changed in last 1-4 years since last high school history books were updated and printed.


> In my experience all printed material is incomplete, obsolete, or one-sided and frequently all three.

Maybe you should stop reading exclusively tabloids...

On a serious note, a middle schooler writing a report on the founding fathers will have plenty of resources available to them in the school library.



Abstract: The upcoming European General Data Protection Regulation (GDPR) prohibits the processing and exploitation of some categories of personal data (health, political orientation, sexual preferences, religious beliefs, ethnic origin, etc.) due to the obvious privacy risks that may be derived from a malicious use of such type of information. These categories are referred to as sensitive personal data. Facebook has been recently fined €1.2M in Spain for collecting, storing and processing sensitive personal data for advertising purposes. This paper quantifies the portion of Facebook users in the European Union (EU) who are labeled with interests linked to sensitive personal data. The results of our study reveal that Facebook labels 73% EU users with sensitive interests. This corresponds to 40% of the overall EU population. We also estimate that a malicious third-party could unveil the identity of Facebook users that have been assigned a sensitive interest at a cost as low as €0.015 per user. Finally, we propose and implement a web browser extension to inform Facebook users of the sensitive interests Facebook has assigned them.


Abstract: Hardware-based mechanisms for software isolation are becoming increasingly popular, but implementing these mechanisms correctly has proved difficult, undermining the root of security. This work introduces an effective way to formally verify important properties of such hardware security mechanisms. In our approach, hardware is developed using a lightweight security-typed hardware description language (HDL) that performs static information flow analysis. We show the practicality of our approach by implementing and verifying a simplified but realistic multi-core prototype of the ARM TrustZone architecture. To make the security-typed HDL expressive enough to verify a realistic processor, we develop new type system features. Our experiments suggest that information flow analysis is efficient, and programmer effort is modest. We also show that information flow constraints are an effective way to detect hardware vulnerabilities, including several found in commercial processors.


Using something like Geocoq [0] to check the proof would be amazing.

[0] http://geocoq.github.io/GeoCoq/


The idea seems interesting but I can't try it, Papr [0] uses Google to login...

[0] https://jhubiostatistics.shinyapps.io/papr


I think Stan uses sampling-based Monte Carlo, whereas PSI is based on static program analysis.


I'm a Stan user and I guess I'm interested to know if this would solve similar problems.


Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: