Variational autoencoders (VAEs) are a popular framework for modeling complex data distributions; they can be efficiently trained via variational inference by maximizing the evidence lower bound (ELBO), at the expense of a gap to the exact (log-)marginal likelihood. While VAEs are commonly used for representation learning, it is unclear why ELBO maximization would yield useful representations, since unregularized maximum likelihood estimation cannot invert the data-generating process. Yet, VAEs often succeed at this task. We seek to elucidate this apparent paradox by studying nonlinear VAEs in the limit of near-deterministic decoders. We first prove that, in this regime, the optimal encoder approximately inverts the decoder---a commonly used but unproven conjecture---which we refer to as self-consistency. Leveraging self-consistency, we show that the ELBO converges to a regularized log-likelihood. This allows VAEs to perform what has recently been termed independent mechanism analysis (IMA): it adds an inductive bias towards decoders with column-orthogonal Jacobians, which helps recovering the true latent factors. The gap between ELBO and log-likelihood is therefore welcome, since it bears unanticipated benefits for nonlinear representation learning. In experiments on synthetic and image data, we show that VAEs uncover the true latent factors when the data generating process satisfies the IMA assumption.

If you're sick of AI being shoved into in more and more consumer electronics, then you're in good company: I'm sick of writing about it. But Google most decidedly is not sick of it. In fact, Google is trying to inject its own AI into every one of its products. That includes both Android and ChromeOS, which are getting a little closer to each other to achieve this end goal. According to the latest Chromium blog post, the Google-developed laptop and desktop system will be "embracing portions of the Android stack, like the Android Linux kernel and Android frameworks, as part of the foundation of ChromeOS."

Artificial Intelligence (AI) is having a transformative effect on a huge range of industries, and the world of media and entertainment is no exception. Creators and machines are continuing to become more intertwined, with creative workflows taking on new shapes as AI-assistance gathers momentum. At a broad level, people are recognising that technology and creativity go hand in hand. Creative professionals are expressing an interest in how AI and machine learning can aid the creative process. And although the discussion about machines replacing humans remains prevalent, the reality is much less dystopian.

As the world's most powerful computer systems begin to embrace artificial intelligence in earnest, using smart algorithms to increase efficiency and speed, the potential damage that a "rogue" AI could cause continues to grow. Facebook engineers were forced to pull the plug on one of the company's AI systems after its bots began communicating with each other in a completely new language which humans simply couldn't decipher. Without being able to understand how or why the bots were communicating, Facebook had no choice but to shut down the system. Facebook's Mark Zuckerberg is a huge proponent of artificial intelligence, and has even clashed with other tech leaders like Elon Musk over his embrace of AI and apparent lack of concern over the theoretical dangers intelligent computers could pose to humanity.