Discrete latent-variable models, while applicable in a variety of settings, can often be difficult to learn. Sampling discrete latent variables can result in high-variance gradient estimators for two primary reasons: 1. branching on the samples within the model, and 2. the lack of a pathwise derivative for the samples. While current state-of-the-art methods employ control-variate schemes for the former and continuous-relaxation methods for the latter, their utility is limited by the complexities of implementing and training effective control-variate schemes and the necessity of evaluating (potentially exponentially) many branch paths in the model. Here, we revisit the reweighted wake-sleep (RWS) (Bornschein and Bengio, 2015) algorithm, and through extensive evaluations, show that it circumvents both these issues, outperforming current state-of-the-art methods in learning discrete latent-variable models. Moreover, we observe that, unlike the importance weighted autoencoder, RWS learns better models and inference networks with increasing numbers of particles, and that its benefits extend to continuous latent-variable models as well. Our results suggest that RWS is a competitive, often preferable, alternative for learning deep generative models.

Frequency-specific patterns of neural activity are traditionally interpreted as sustained rhythmic oscillations, and related to cognitive mechanisms such as attention, high level visual processing or motor control. While alpha waves (8-12 Hz) are known to closely resemble short sinusoids, and thus are revealed by Fourier analysis or wavelet transforms, there is an evolving debate that electromagnetic neural signals are composed of more complex waveforms that cannot be analyzed by linear filters and traditional signal representations. In this paper, we propose to learn dedicated representations of such recordings using a multivariate convolutional sparse coding (CSC) algorithm. Applied to electroencephalography (EEG) or magnetoencephalography (MEG) data, this method is able to learn not only prototypical temporal waveforms, but also associated spatial patterns so their origin can be localized in the brain. Our algorithm is based on alternated minimization and a greedy coordinate descent solver that leads to state-of-the-art running time on long time series. To demonstrate the implications of this method, we apply it to MEG data and show that it is able to recover biological artifacts. More remarkably, our approach also reveals the presence of non-sinusoidal mu-shaped patterns, along with their topographic maps related to the somatosensory cortex.

Tesla Inc has flown six planes full of robots and equipment from Europe to California in an unusual, high-stakes effort to speed up battery production for its Model 3 electric sedan, people familiar with the matter told Reuters this week. Transporting equipment for a production line by air is costly and hardly ever done in the automotive industry, and the move underscores Tesla Chief Executive Elon Musk's urgency to get a grip on manufacturing problems that have hobbled the launch of the high-volume Model 3 and pushed Tesla's finances deep into the red. 'As usual with Tesla, everything is being done in a massive hurry and money seems to be no obstacle,' said one of the two sources. Tesla on Friday declined to comment on whether it has shipped in any new production equipment from Europe. Transporting equipment for a production line by air is costly and hardly ever done in the automotive industry, and the move underscores Tesla Chief Executive Elon Musk's urgency to get a grip on manufacturing problems Investors are closely watching Tesla and its high-profile, often brash CEO to see if the upstart electric vehicle maker can pull off high-volume production of the Model 3, a car with the potential to catapult the niche automaker to a mass producer and assure its financial stability.

The industrial revolution was a major turning point in history. Until the advent of steam engine technology, economic growth and wealth-creation were stagnant. After 1800, economic growth statistics accelerated, and they have continued to do so ever since. This happened initially in industrial nations across Europe and North America, but everywhere else soon followed. In the centuries since we have witnessed subsequent revolutions in industrial technology, from chemistry and electricity in the 19th century to computer technology in the 20th. Today artificial intelligence (AI) is enabling the next industrial revolution, but this one is very different to what has come before.

SiFive does a quarterly series of tech talks, not necessarily directly to do with SiFive or even RISC-V. For example, last quarter it was Paul Kocher (and if you don't know that name, you need to go and read my post about that talk Paul Kocher: Differential Power Analysis and Spectre). This quarter it was Krste Asanović on Accelerating AI: Past, Present, and Future. This post will cover the past. The present and future have to wait (good title for a movie?).

A couple of weeks ago, I pointed out three compelling Artificial Intelligence (AI) stocks you may not have even known about. There are certainly more than three such artificial intelligence prospects though. So, today I'd like to introduce you to three more possibilities we just didn't have the time or room to explore with the first go-around. Like last time, the goal isn't to state the obvious. Most investors fully understand that Nvidia Corporation (NASDAQ:NVDA) makes the hardware that powers most AI applications while International Business Machines Corporation (NYSE:IBM) is arguably doing the most to put artificial intelligence to practical use.

Provided by AFP Your next job interview may be with this... You have a telephone interview for your dream job, and you're feeling nervous. You make yourself a cup of tea as you wait for the phone to ring, and you count to three before picking up. Now imagine that your interviewer is a robot named Vera. Russian startup Stafory co-founder Alexei Kostarev says Robot Vera, which his company developed, is driven by artificial intelligence (AI) algorithms.

The hysteria about the future of artificial intelligence (AI) is everywhere. There seems to be no shortage of sensationalist news about how AI could cure diseases, accelerate human innovation and improve human creativity. Just looking at the media headlines, you might think that we are already living in a future where AI has infiltrated every aspect of society. While it is undeniable that AI has opened up a wealth of promising opportunities, it has also led to the emergence of a mindset that can be best described as "AI solutionism". This is the philosophy that, given enough data, machine learning algorithms can solve all of humanity's problems.

It is a device normally used to check the weather forecast or order more toilet paper. But Alexa can now summon a prayer to say grace before your evening meal or give thanks to God before going to bed at night. Amazon's virtual assistant has been picked up by the Church of England to make being a practising Christian easier. In response to a simple command, its app gets Alexa to read out the Lord's Prayer or the Ten Commandments, for those concerned about wrongdoing. Along with other Google searches, users can ask how to become a Christian, how to pray and who God is.

Artificial intelligence and machine learning saw a significant spike of attention in the past few years – whether it's through partnerships, acquisitions, or in-house developments. The largest financial institutions in the US have been involved in one way or another in bringing artificial intelligence into operations and customer-facing functions. A recent study of 34 major banks across several geographies (US, EU, Singapore, Africa, Australia, India) by MEDICI Team found that 27 out of these 34 banks have implemented AI in their front-office functions in form of a chatbot, virtual assistant, and digital advisor. Some of the most prominent banks in this space across regions are Bank of America, OCBC, ABN Amro, YES BANK, etc. While front-office applications have certainly seen a higher intensity, scope, and adoption, the AI strategy in the US banking industry, in reality, is far more diverse.