Ourcontributionistoextend theidentifiability theory ofdeepgenerativemodels for a very broad class of structured models.

Ourcontributionistoextend theidentifiability theory ofdeepgenerativemodels for a very broad class of structured models.

Plus: Apple's Lockdown mode keeps the FBI out of a reporter's phone, Elon Musk's Starlink cuts off Russian forces, and more. An analysis by WIRED this week found that ICE and CBP's face recognition app Mobile Fortify, which is being used to identify people across the United States, isn't actually designed to verify who people are and was only approved for Department of Homeland Security use by relaxing some of the agency's own privacy rules. WIRED took a close look at highly militarized ICE and CBP units that use extreme tactics typically seen only in active combat. Two agents involved in the shooting deaths of US citizens in Minneapolis are reportedly members of these paramilitary units. And a new report from the Public Service Alliance this week found that data brokers can fuel violence against public servants, who are facing more and more threats but have few ways to protect their personal information under state privacy laws.

Classical causal structure learning algorithms often assume no latent confounders. However, it is usually impossible to enumerate and measure all task-related variables in real-world scenarios. Neglecting latent confounders may lead to spurious correlations among observed variables.

We show that PTFs of arbitrary constant degree can be testably learned up to excess errorε > 0 in time npoly(1/ε). This qualitatively matches the best known guarantees in the agnostic model. Our results build on a connection between testable learning andfooling.