Statistical methods for network data have received a lot of attention because of the wideranging applications of network analysis. There is now a large body of work on methods and models for networks, including the stochastic block model (SBM) [Holland et al., 1983], the degree-corrected stochastic block model (DCSBM) [Karrer and Newman, 2011], and the latent space model [Hoff et al., 2002], to name a few. While this gives the practitioner plenty of choices, there is a lot less work on the crucial question of how to select the best model for the data, as well as how to choose tuning parameters for the selected model, which is often necessary in order to fit it. In some specific problems, progress has been made recently, for instance, in the much-studied problem of community detection. Community detection is the problem of clustering network nodes into groups, and most of the methods proposed over the last twenty years or so require the number of communities K as input.

In this paper, we propose and study opportunistic bandits - a new variant of bandits where the regret of pulling a suboptimal arm varies under different environmental conditions, such as network load or produce price. When the load/price is low, so is the cost/regret of pulling a suboptimal arm (e.g., trying a suboptimal network configuration). Therefore, intuitively, we could explore more when the load is low and exploit more when the load is high. Inspired by this intuition, we propose an Adaptive Upper-Confidence-Bound (AdaUCB) algorithm to adaptively balance the exploration-exploitation tradeoff for opportunistic bandits. We prove that AdaUCB achieves $O(\log T)$ regret with a smaller coefficient than the traditional UCB algorithm. Furthermore, AdaUCB achieves $O(1)$ regret when the exploration cost is zero if the load level is below a certain threshold. Last, based on both synthetic data and real-world traces, experimental results show that AdaUCB significantly outperforms other bandit algorithms, such as UCB and TS (Thompson Sampling), under large load fluctuations.

Most distributed machine learning systems nowadays, including TensorFlow and CNTK, are built in a centralized fashion. One bottleneck of centralized algorithms lies on high communication cost on the central node. Motivated by this, we ask, can decentralized algorithms be faster than its centralized counterpart? Although decentralized PSGD (D-PSGD) algorithms have been studied by the control community, existing analysis and theory do not show any advantage over centralized PSGD (C-PSGD) algorithms, simply assuming the application scenario where only the decentralized network is available. In this paper, we study a D-PSGD algorithm and provide the first theoretical analysis that indicates a regime in which decentralized algorithms might outperform centralized algorithms for distributed stochastic gradient descent. This is because D-PSGD has comparable total computational complexities to C-PSGD but requires much less communication cost on the busiest node. We further conduct an empirical study to validate our theoretical analysis across multiple frameworks (CNTK and Torch), different network configurations, and computation platforms up to 112 GPUs. On network configurations with low bandwidth or high latency, D-PSGD can be up to one order of magnitude faster than its well-optimized centralized counterparts.

A drone is flown during a property inspection following Hurricane Harvey in Houston. The mass destruction brought on by Harvey has been a seminal moment for drone operators, proving that they can effectively map flooding, locate people in need of rescue and verify damage to speed insurance claims. A drone is flown during a property inspection following Hurricane Harvey in Houston. The mass destruction brought on by Harvey has been a seminal moment for drone operators, proving that they can effectively map flooding, locate people in need of rescue and verify damage to speed insurance claims. Tropical Storm Harvey disrupted at least 17 emergency call centers and 320 cellular sites, and it caused outages for more than 148,000 Internet, TV, and phone customers, according to the Federal Communications Commission.

SoftBank Group Corp. is getting ready to return to the overseas debt market for the second time in two months, after issuing $4.5 billion in subordinated notes in July. The Tokyo-based company headed by billionaire founder Masayoshi Son is accelerating deal-making around the world, and has been in discussions to combine its U.S. wireless operator Sprint Corp. with a potential partner. The SoftBank Vision Fund has also kicked off investments, with deals ranging from ride-sharing, co-working and robotics to agriculture, cancer detection and autonomous driving. SoftBank will hold investor calls on Sept. 8 for potential U.S. dollar, euro note sales, according to a person familiar with this offering, who is not authorized to speak publicly and asked not to be identified. The Japanese internet giant has mandated Deutsche Bank AG, Citigroup Inc., Goldman Sachs Group Inc. and Morgan Stanley for the deal.

As part of the Scyfer acquisition announcement, Qualcomm laid out its research focus for artificial intelligence and machine learning, which is expected to play a larger role in upcoming generations of gadgets. Qualcomm already delivers artificial intelligence features on its latest smartphone chips, including technologies that help smartphone cameras recognize objects and detect surroundings to aid in better photo composition. Prototypes for President Donald Trump's border wall may be completed by Halloween. Here are three things Snap must show in its next earnings report to satisfy investors:(August 11, 2017) (Sign up for our free video newsletter here http://bit.ly/2n6VKPR) Here are three things Snap must show in its next earnings report to satisfy investors:(August 11, 2017) (Sign up for our free video newsletter here http://bit.ly/2n6VKPR) Miko Branch was deep asleep when her sister Titi woke her up to celebrate.

Let's see how to build a Telegram Bot which checks the emotions in your text with a quantitative approach through sentiment analysis. As already said, the weapon of choice is a Telegram Bot for the APIs ease of use, but any other programmable messenger system will work. A lot of companies offer a good text analysis API (IBM Watson, Google, Api.ai), but this time I decided to use Indico for the ease of use, the nice dashboard and mostly for the node module we can easily integrate in our webtask. We have a running bot, a running webtask which is not getting any message so far () and our AI ready to analyse our emotions and sentiments ().

You might not have, but Alexa did. Voice assistants have been successfully hijacked using sounds above the range of human hearing. Once in, hackers were able to make phone calls, post on social media and disconnect wireless services, among other things. Assistants falling for the ploy included Amazon Alexa, Apple's Siri, Google Now, Samsung S Voice, Microsoft Cortana and Huawei HiVoice, as well as some voice control systems used in cars. The hack was created by Guoming Zhang, Chen Yan and their team at Zhejiang University in China.

When someone gives you a good news and you reply with "great" maybe you are not exactly transmitting the happiest of the feelings, but you are happy, really. Lucky us, we live in the age of Artificial Intelligence and writing some code has never been so easy! Let's see how to build a Telegram Bot which checks the emotions in your text with a quantitative approach through sentiment analysis. As already said, the weapon of choice is a Telegram Bot for the APIs ease of use, but any other programmable messenger system will work. What really makes all interesting is the chance to build all this in a serverless architecture: thanks to Webtask.io, which offers an amazing environment, security (powered by Auth0) and serverless endpoints, our server side core will receive the message, send it to the our Text Analysis engine and prettify the answer in order to reply eventually with a nicely formatted response.

The team has developed an app called PupilScreen that uses video and a smartphone's camera flash to record and calculate how the pupils respond to light. Assessing head trauma due to, for example, sports injuries or a car crash is typically done with either a pupilometer -- rarely found outside of hospitals -- or a mix of subjective evaluations like balancing, repeating a list of words or visually examining a pupil's response with a flashlight. To create PupilScreen and provide an objective assessment of potential head trauma, the researchers used deep learning tools to train a neural network how to find the pupil of the eye and track how it responds to a flash of light over the course of three seconds. A smartphone camera records the three second video and the light is provided by the camera's flash.