While it is believed that denoising is not always necessary in many big data applications, we show in this paper that denoising is helpful in urban traffic analysis by applying the method of bounded total variation denoising to the urban road traffic prediction and clustering problem. We propose two easy-to-implement methods to estimate the noise strength parameter in the denoising algorithm, and apply the denoising algorithm to GPS-based traffic data from Beijing taxi system. For the traffic prediction problem, we combine neural network and history matching method for roads randomly chosen from an urban area of Beijing. Numerical experiments show that the predicting accuracy is improved significantly by applying the proposed bounded total variation denoising algorithm. We also test the algorithm on clustering problem, where a recently developed clustering analysis method is applied to more than one hundred urban road segments in Beijing based on their velocity profiles. Better clustering result is obtained after denoising.

The increasing availability of electrocardiogram (ECG) data has motivated the use of data-driven models for automating various clinical tasks based on ECG data. The development of subject-specific models are limited by the cost and difficulty of obtaining sufficient training data for each individual. The alternative of population model, however, faces challenges caused by the significant inter-subject variations within the ECG data. We address this challenge by investigating for the first time the problem of learning representations for clinically-informative variables while disentangling other factors of variations within the ECG data. In this work, we present a conditional variational autoencoder (VAE) to extract the subject-specific adjustment to the ECG data, conditioned on task-specific representations learned from a deterministic encoder. To encourage the representation for inter-subject variations to be independent from the task-specific representation, maximum mean discrepancy is used to match all the moments between the distributions learned by the VAE conditioning on the code from the deterministic encoder. The learning of the task-specific representation is regularized by a weak supervision in the form of contrastive regularization. We apply the proposed method to a novel yet important clinical task of classifying the origin of ventricular tachycardia (VT) into pre-defined segments, demonstrating the efficacy of the proposed method against the standard VAE.

The rapid development of big data techniques has offered great opportunities to develop smart city services in public safety, transportation management, city planning, etc. Meanwhile, the smart city development levels of different cities are still unbalanced. For a large of number of cities which just start development, the governments will face a critical cold-start problem, 'how to develop a new smart city service suffering from data scarcity?'. To address this problem, transfer learning is recently leveraged to accelerate the smart city development, which we term the urban transfer learning paradigm. This article investigates the common process of urban transfer learning, aiming to provide city governors and relevant practitioners with guidelines of applying this novel learning paradigm. Our guidelines include common transfer strategies to take, general steps to follow, and case studies to refer. We also summarize a few future research opportunities in urban transfer learning, and expect this article can attract more researchers into this promising area.

In recent years there has been a sharp rise in networking applications, in which significant events need to be classified but only a few training instances are available. These are known as cases of one-shot learning. Examples include analyzing network traffic under zero-day attacks, and computer vision tasks by sensor networks deployed in the field. To handle this challenging task, organizations often use human analysts to classify events under high uncertainty. Existing algorithms use a threshold-based mechanism to decide whether to classify an object automatically or send it to an analyst for deeper inspection. However, this approach leads to a significant waste of resources since it does not take the practical temporal constraints of system resources into account. Our contribution is threefold. First, we develop a novel Deep Reinforcement One-shot Learning (DeROL) framework to address this challenge. The basic idea of the DeROL algorithm is to train a deep-Q network to obtain a policy which is oblivious to the unseen classes in the testing data. Then, in real-time, DeROL maps the current state of the one-shot learning process to operational actions based on the trained deep-Q network, to maximize the objective function. Second, we develop the first open-source software for practical artificially intelligent one-shot classification systems with limited resources for the benefit of researchers in related fields. Third, we present an extensive experimental study using the OMNIGLOT dataset for computer vision tasks and the UNSW-NB15 dataset for intrusion detection tasks that demonstrates the versatility and efficiency of the DeROL framework.

A decade ago, a group of Internet companies, civil society organizations, academics, and investors launched the Global Network Initiative (GNI), a collaborative effort to promote free expression and protect user privacy on the Internet. Google helped lead this effort and a parallel project devoted to developing a human rights framework for the Internet. In 2010, Google further demonstrated its leadership by making a principled decision to withdraw its search-engine services from China. In a very public way, the company acknowledged the inherent contradiction between Chinese Internet censorship and Google's commitments to its users and the GNI to promote free expression. It was thus disturbing to read recent reports suggesting that Google now is seriously considering re-entering the Chinese market and succumbing to Chinese censorship in exchange for commercial opportunity.

A mere sprinkling of autonomous vehicles exist in a few dozen cities today. And none of them -- at least not yet -- have been deployed as a true commercial enterprise. While the bulk of this nascent industry fixates on the system of sensors, maps and AI necessary for vehicles to drive without a human behind the wheel, the founders of startup RideOS are directing their efforts to the day when fleets of self-driving cars hit the streets. It's there, where human-driven and automated vehicles will be forced to mingle, that RideOS co-founders Chris Blumenberg and Justin Ho see opportunity. The company, which has existed for all of 12 months, has raised $25 million in a Series B funding round led by Next47, the venture arm of Siemens. Sequoia, an existing investor, and Singapore-based ST Ventures, also participated in the round.

The Apollo 3.0 Launch Event in Mountain View last night highlighted the strides made by the open-source self-driving car project backed by Chinese Internet giant Baidu. The project's leaders announced a new collection of low-speed driving capabilities, such as delivery vehicle driving and self-parking, alongside a wide array of sensors that independent groups can now connect with Apollo software. Apollo is becoming increasingly prominent as a foundation for automotive and technology companies that want to develop autonomous vehicles, but do not want to develop the entire software stack themselves. Two other companies, NVIDIA and Tier IV, also support software stacks that third-parties can use to develop self-driving cars. NVIDIA's DRIVE software platform is tightly coupled to their DRIVE line of computational units.

Apple announced stellar quarterly earnings; investors liked them; the stock rose; and Apple became the first US company to surpass $1 trillion in market value. In our love for big numbers, that made it a big story. Zachary Karabell is a WIRED contributor and president of River Twice Research. Never mind that if you adjust for inflation and go global, Apple isn't actually the first trillion-dollar company. PetroChina, the state-owned Chinese oil company, hit that number more than a decade ago, and in adjusted terms, Standard Oil did a century ago.

While nuances separate their individual capabilities, they all take roughly the same form: a human-like voice embedded into a smart speaker, mobile phone, automobile, or similar piece of hardware. Japanese messaging giant Line, however, wants to give virtual assistants a more human form. The company this week opened preorders for the latest iteration of the Gatebox virtual home robot, a holographic character that is designed to provide companionship to its owner. The Gatebox is a little table lamp-sized glass case that uses projections and sensors to create a life-like character, called Hikari Azuma, that the user can interact with. By way of a quick recap, the Gatebox was originally developed by a Japanese firm called Vinclu, which launched a limited run of 300 units in 2016, priced at the equivalent of around $2,670 each. It followed up a year later with another run of 39 units before Line acquired a majority stake in the company.

President Xi Jinping's "Made in China 2025" strategy, unveiled in 2015 and now thrust back into the limelight by President Trump's bellicose stance on trade, holds three important lesson for global industry. This should be a non-controversial statement, but it is not. Economists often mock "the manufacturing fetish" and argue there is no reason to consider manufacturing a better driver of economic growth than any other sector. As economies get richer, they tend to shift from agriculture to industry, and then to services. Manufacturing accounted for nearly 30% of the U.S. economy in the 1950s; it was still 20% in the 1980s; today it accounts for just 11%.