DL4J is a pretty awesome open source project that works with Spark and Hadoop. Deep Learning 4J also works as a YARN app! It includes Text, NLP, Canova Vectorization Lib for ML, Scientific computing for the JVM, distributed with clusters, and works with CUDA GPU kernels. DL4J is used for anomaly detection (fraud detection), recommender systems, predictive analytics with logs and image recognition. In a related open source project, Skymind built a numerical computing library ND4J, or n-dimensional arrays for Java, essentially porting Numpy to the JVM.

Three Kansas City startups will receive a combined 59,000 from the Mozilla Gigabit Community Fund to expand and develop programs that promote innovation in the classroom. KC Social Innovation Center, PlanIT Impact and Pennez were awarded money for using Kansas City's gigabit internet to create new ways to learn. The KC Social Innovation Center will give students real-world experience in the emerging'Internet of Things' industry. The internet of things is the network of physical objects -- devices, vehicles, buildings -- containing software, sensors and internet connectivity that enable them to collect and exchange data. PlanIT Impact is a tool that provides architects, planners and designers with information on how a building or site will utilize energy, emit greenhouse gases and perform in other ways by using open data to create interactive 3D models.

This week's economic calendar is pretty light. Market participants will be looking to an early getaway for the long weekend. While there will be plenty of entertaining FedSpeak, I expect a different topic to be at the fore. The news was pretty good, but the stock market was not. In my last WTWA, I predicted that the punditry would be asking whether it was "springtime for housing". That was the recurring topic as housing news was reported on several different days and garnered plenty of discussion. Competition came from the Fed Minutes, some dramatic earnings reports, and the election race. I always start my personal review of the week by looking at this great chart from Doug Short.

Four sister robots built by NASA could be pioneers in the colonization of Mars, part of an advance construction team that sets up a habitat for more fragile human explorers. But first they're finding new homes on Earth and engineers to hone their skills.

This week on TechRepublic's Business Technology Weekly podcast, hosts Dan Patterson and Bill Detwiler discuss how swarm AI won the Kentucky derby, and the real world, practical impact of artificial intelligence. Headlines: Swarm AI predicts the 2016 Kentucky Derby Hope Reese Big news in the AI world this week! HOME NEWS Baidu to Shift to AI After Government Probe Baidu is planning to switch toward developing artificial intelligence after a government probe that affected its core business. BERLIN, GERMANY - SEPTEMBER 04: Visitors look at smartphones at the Lenovo stand at the 2015 IFA consumer electronics and appliances trade fair on September 4, 2015 in Berlin, Germany. The PC maker posted its first loss in six years in 2015.

This article is for software developers--particularly those coming from a Ruby or Java language background--who are facing their first machine learning implementation. I was recently given the assignment to create an RSS feed categorization subsystem for a client. The goal was to read dozens or even hundreds of RSS feeds and automatically categorize their many articles into one of dozens of predefined subject areas. The content, navigation, and search functionality of the client website would be driven by the results of this daily automated feed retrieval and categorization. The client suggested using machine learning, perhaps with Apache Mahout and Hadoop, as she had recently read articles about those technologies. Her development team and ours, however, are fluent in Ruby rather than Java technology. This article describes the technical journey, learning process, and ultimate implementation of a solution. My first question was, "what exactly is machine learning?"

Data Science Middle East (#DSME) in partnership with PAPIs are excited to announce this 2-day hands-on Machine Learning Workshop. Most Machine Learning courses are given from the perspective of a researcher/academic and focus on the theory and mathematics of the machine learning models. This workshop takes the perspective of learning by working on real machine learning problems using open source tools and platforms. We'll go all the way from data preparation to the integration of predictive models in applications and their deployment in production. "Just like development where you don't need to know a thing about computability or big-O notation to write code and ship useful and reliable software, you can work machine learning problems end-to-end without a background in statistics, probability and linear algebra."

It's hard to make a right turn down San Francisco's crowded streets these days without running into a story about bots, conversational commerce, or conversational commerce bots. Bots have reached kale levels of hype. But amidst the noise, there's something real going on here -- we just have to decipher it. You may have heard of conversational commerce -- it's a catchall term for a future of technology driven by messaging (and voice) interactions that transcend current communications modalities. It's a convenient moniker but also confusing because there isn't one trend to follow.

One of the major issues in stochastic gradient descent (SGD) methods is how to choose an appropriate step size while running the algorithm. Since the traditional line search technique does not apply for stochastic optimization algorithms, the common practice in SGD is either to use a diminishing step size, or to tune a fixed step size by hand, which can be time consuming in practice. In this paper, we propose to use the Barzilai-Borwein (BB) method to automatically compute step sizes for SGD and its variant: stochastic variance reduced gradient (SVRG) method, which leads to two algorithms: SGD-BB and SVRG-BB. We prove that SVRG-BB converges linearly for strongly convex objective functions. As a by-product, we prove the linear convergence result of SVRG with Option I proposed in [10], whose convergence result is missing in the literature. Numerical experiments on standard data sets show that the performance of SGD-BB and SVRG-BB is comparable to and sometimes even better than SGD and SVRG with best-tuned step sizes, and is superior to some advanced SGD variants.

Recent work in distance metric learning has focused on learning transformations of data that best align with provided sets of pairwise similarity and dissimilarity constraints. The learned transformations lead to improved retrieval, classification, and clustering algorithms due to the better adapted distance or similarity measures. Here, we introduce the problem of learning these transformations when the underlying constraint generation process is nonstationary. This nonstationarity can be due to changes in either the ground-truth clustering used to generate constraints or changes to the feature subspaces in which the class structure is apparent. We propose and evaluate COMID-SADL, an adaptive, online approach for learning and tracking optimal metrics as they change over time that is highly robust to a variety of nonstationary behaviors in the changing metric. We demonstrate COMID-SADL on both real and synthetic data sets and show significant performance improvements relative to previously proposed batch and online distance metric learning algorithms.