In this paper we introduce the ice-start problem, i.e., the challenge of deploying machine learning models when only little or no training data is initially available, and acquiring each feature element of data is associated with costs. This setting is representative for the real-world machine learning applications. For instance, in the health-care domain, when training an AI system for predicting patient metrics from lab tests, obtaining every single measurement comes with a high cost. Active learning, where only the label is associated with a cost does not apply to such problem, because performing all possible lab tests to acquire a new training datum would be costly, as well as unnecessary due to redundancy. We propose Icebreaker, a principled framework to approach the ice-start problem. Icebreaker uses a full Bayesian Deep Latent Gaussian Model (BELGAM) with a novel inference method. Our proposed method combines recent advances in amortized inference and stochastic gradient MCMC to enable fast and accurate posterior inference. By utilizing BELGAM's ability to fully quantify model uncertainty, we also propose two information acquisition functions for imputation and active prediction problems. We demonstrate that BELGAM performs significantly better than the previous VAE (Variational autoencoder) based models, when the data set size is small, using both machine learning benchmarks and real-world recommender systems and health-care applications. Moreover, based on BELGAM, Icebreaker further improves the performance and demonstrate the ability to use minimum amount of the training data to obtain the highest test time performance.

For over 50 years, researchers have been trying to teach computers to read music notation, referred to as Optical Music Recognition (OMR). However, this field is still difficult to access for new researchers, especially those without a significant musical background: few introductory materials are available, and furthermore the field has struggled with defining itself and building a shared terminology. In this tutorial, we address these shortcomings by (1) providing a robust definition of OMR and its relationship to related fields, (2) analyzing how OMR inverts the music encoding process to recover the musical notation and the musical semantics from documents, (3) proposing a taxonomy of OMR, with most notably a novel taxonomy of applications. Additionally, we discuss how deep learning affects modern OMR research, as opposed to the traditional pipeline. Based on this work, the reader should be able to attain a basic understanding of OMR: its objectives, its inherent structure, its relationship to other fields, the state of the art, and the research opportunities it affords.

LOS ANGELES, Aug. 13 (Xinhua) -- Researchers at University of Utah Health are using machine learning to draw links between genetic controls that …

HPCwire: Walk us through the program, give us a sense of what these AI and science town halls are all about and what they are trying to accomplish? RS: If you remember back in 2007, we had three town hall meetings – at Argonne, Berkeley and Oak Ridge – that launched the whole DOE Exascale project and so forth. At that time the idea was to get people together and ask them, for exascale, what if we could build these faster machines, what would you do with them. It was a way to get people thinking about the possibility of that and of course it took long time to get the exascale computing program going. With these town halls we are kind of asking a variation on that question. Now we're asking the question of what's the opportunity for AI in science or the application of science, particularly in the context of DOE, but more broadly because DOE's got a lot of collaborations with NIH and other agencies. So really asking the fundamental question of what do we have to do in the AI space to make it relevant for science. The point of the town halls – three in the labs and one in Washington in October – is go get people thinking about what opportunities there are in different scientific domains for breakthrough science that can be accomplished by leveraging AI and working AI into simulation, and bringing AI into big data, bringing AI to the facility and so forth. So that's the concept; it's really to get the community moving.

Israeli pedestrian detection start-up, Viziblezone, reports it has reached a major milestone in the development of its detector technology. The company claims that its prototype system has proven that it can detect pedestrians even hidden behind objects at distances of up to 150 metres. According to Viziblezone, while many technologies to mitigate vehicle-to-vehicle accidents have been developed in recent years, there remains a significant lack of vehicle-to-pedestrian accident prevention systems. Meanwhile, with the growth of autonomously driven vehicles, and the expansion of technologies such as robo-taxis, the risks to pedestrians are increasing exponentially at a rate that existing vehicle sensor systems can't effectively address, the company notes In June 2019, the World Health Organisation reported that in 2018, more than 1.5 million people were killed, and more than 50 million injured in road accidents. Of these casualties, more than 50 per cent were pedestrians and cyclists – a number that tragically continues to grow.

A few years ago, consulting giant McKinsey predicted that artificial intelligence and automation would eliminate a 73 million jobs by 2030. That's a scary number of jobs, and a prediction that led many professionals to evaluate their own skills and research ways to future-proof their careers from the coming automation/AI job apocalypse. In the absence of an elder expert giving you one word of advice, like "Plastics!" what can an IT professional (or a new graduate) do to ensure that the robots don't come to take your job? That's been a key question over the past few years, and one without that simple one-word answer. Experts have said that jobs focused on implementing and managing artificial intelligence and automation would be great avenues for job-seekers to pursue.

Google AI researchers working with the ALS Therapy Development Institute today shared details about Project Euphonia, a speech-to-text transcription service for people with speaking impairments. They also say their approach can improve automatic speech recognition for people with non-native English accents as well. People with amyotrophic lateral sclerosis (ALS) often have slurred speech, but existing AI systems are typically trained on voice data without any affliction or accent. The new approach is successful primarily due to the introduction of small amounts of data that represents people with accents and ALS. "We show that 71% of the improvement comes from only 5 minutes of training data," according to a paper published on arXiv July 31 titled "Personalizing ASR for Dysarthric and Accented Speech with Limited Data."

Chinese search engine Sogou is creating artificial-intelligence lookalikes to read popular novels in authors' voices. It announced "lifelike" avatars of Chinese authors Yue Guan and Bu Xin Tian Shang Diao Xian Bing - created from video recordings - at the China Online Literature conference. Last year, Sogou launched two AI newsreaders, which are still used by the government's Xinhua news agency. Appetite for audiobooks in China is on the rise, mirroring trends in the West. Chinese think tank iiMedia expects the market to more than double between 2016 and 2020, to 7.8bn Chinese yuan (£900m) a year. It is now a simple process to use text-to-speech technology to quickly generate an audio version of a book, using digitised, synthetic voices.

Utilities and energy retailers from across the globe will gather at the exclusive Bidgely Engage 2019 conference under the banner of'Unlock the Power of UtilityAI ' this September 11-13 in Napa, Calif. Engage 2019 is utility artificial intelligence (AI) leader Bidgely's third annual event that brings together utilities, AI experts and tech leaders to discuss trends, best practices and lessons learned in applied AI for the energy industry, as well as to enjoy networking in California's legendary wine country. This press release features multimedia. "For Engage, we pull in industry luminaries and tech leaders from outside the energy space to learn from their AI journeys and to explore how AI and machine learning advancements specifically for energy is delivering compounding benefits to utilities around the world," said Bidgely CMO Gautam Aggarwal. The shift of AI becoming mainstream in the energy industry was recently cited in a report by Navigant Research, covering how the future of utilities will be driven by the emerging disciplines of machine learning and artificial intelligence.

Most financial institutions know it's critical to manage the ever-increasing amounts of accessible data, but many miss the potential in using that data in innovative ways. Financial institutions have a plethora of data they can access, either through their own systems or through public sources. However, many can't -- or won't -- exploit the large volumes of data, particularly the "owned" data that an organization holds about customers. This kind of data is typically called customer relationship management data, such as the purchase history tied to app installs, email addresses and postal addresses. Though financial institutions maintain and collect massive volumes of data, many firms are restricted from fully using that data because they are required to comply with stringent regulations around what can and cannot be done with customer data. Such major regulatory changes include the Dodd-Frank Act in the United States, Europe's Markets in Financial Instruments Directive II and the General Data Protection Regulation -- all of which affect banks with a global presence.