Daily Dish 020119

AAAI-19 Invited Talk: Ian Goodfellow (Google AI) [vimeo]  Adversarial Machine Learning Until about 2013, most researchers studying machine learning for artificial intelligence all worked on a common goal: get machine learning to work for AI-scale tasks. Now that supervised learning works, there is a Cambrian explosion of new research directions: making machine learning secure, making machine learning private, getting machine learning to work for new tasks, reducing the dependence on large amounts of labeled data, and so on. In this talk I survey how adversarial techniques in machine learning are involved in several of these new research frontiers.

New York Insurers Can Evaluate Your Social Media UseIf They Can Prove Why It’s Needed [WSJ] Under guidance released earlier this month, life insurers will have to use statistical and actuarial analysis to determine that any algorithms and data are free of bias against racial minorities and other groups protected by law.  (By whose measure and review -ED …  The department also learned that some vendors are pitching data to insurers that includes “condition or type of an applicant’s electronic devices” and “how the consumer appears in a photograph,” it said in material posted on its website.  Among the actions suggested to avoid the AI’s scissor “Visit the gym with a phone linked to a location-tracking service. If you visit the bar, leave your phone at home.” 

White-Collar Robots Are Coming for Jobs  [WSJ] This new version of the old disruptive duo—automation and globalization—will lower the headcount in many service-sector occupations. The “globotics” transformation will not be gentle. Given the rapacious progress of digital technology, these changes will disorder professional and service-sector jobs radically faster than globalization disrupted the manufacturing sector in the 20th century and the agricultural sector in the 19th century. <—–

#CPDP2019 including New Guidelines on Artificial Intelligence and Data Protection On the occasion of Data Protection Day on 28 January, the Consultative Committee of the Convention for the Protection of Individuals with regard to the Processing of Personal Data (Convention 108) has published Guidelines on Artificial Intelligence and Data Protection.  The guidelines aim to assist policy makers, artificial intelligence (AI) developers, manufacturers and service providers in ensuring that AI applications do not undermine the right to data protection

transmediale.de  live streams

Apple shows Facebook who has the power in an app dispute [NYT]  Apple also briefly demonstrated its power on Thursday with another Silicon Valley giant, Google. Like Facebook, Google had violated Apple’s rules by publicly distributing an app, Screenwise Meter, through a special Apple developer program. The internet search company said some of its internal apps that run on iPhone software were temporarily disrupted.  (this is the kind of stuff you would expect the FTC to be doing, instead we have the CEO of Apple acting like the FTC should – ED)

A deep fake act has been proposed by Senator Sasse, The Malicious Deep Fake Prohibition Act   (PDF) Under the bill [s.3805] , it would be a federal felony for individuals to:

(1) create, with the intent to distribute, a deep fake with the intent that the distribution of the deep fake would facilitate criminal or tortious conduct under Federal, State, local, or Tribal law; or

(2) distribute an audiovisual record with— (A) actual knowledge that the audiovisual record is a deep fake; and (B) the intent that the distribution of the audiovisual record would facilitate criminal or tortious conduct under Federal, State, local, or Tribal law.

How Do You Count Every Solar Panel in the U.S.? Machine Learning and a Billion Satellite Images  The DeepSolar Project, developed by engineers and computer scientists at Stanford University, is a machine learning framework that analyzes a dataset of satellite images in order to identify the size and location of installed solar panels. To accurately count the panels, the DeepSolar team used a machine learning algorithm to analyze more than a billion high-resolution satellite images. The algorithm identified what the team believes to be almost every solar power installation across the contiguous 48 states.


Privacy is Power