The network Lasso is a recently proposed convex optimization method for machine learning from massive network structured datasets, i.e., big data over networks. It is a variant of the well-known least absolute shrinkage and selection operator (Lasso), which is underlying many methods in learning and signal processing involving sparse models. Highly scalable implementations of the network Lasso can be obtained by state-of-the art proximal methods, e.g., the alternating direction method of multipliers (ADMM). By generalizing the concept of the compatibility condition put forward by van de Geer and Buehlmann as a powerful tool for the analysis of plain Lasso, we derive a sufficient condition, i.e., the network compatibility condition, on the underlying network topology such that network Lasso accurately learns a clustered underlying graph signal. This network compatibility condition relates the location of the sampled nodes with the clustering structure of the network. In particular, the NCC informs the choice of which nodes to sample, or in machine learning terms, which data points provide most information if labeled.

We consider the problem of maximizing a non-monotone DR-submodular function subject to a cardinality constraint. Diminishing returns (DR) submodularity is a generalization of the diminishing returns property for functions defined over the integer lattice. This generalization can be used to solve many machine learning or combinatorial optimization problems such as optimal budget allocation, revenue maximization, etc. In this work we propose the first polynomial-time approximation algorithms for non-monotone constrained maximization. We implement our algorithms for a revenue maximization problem with a real-world dataset to check their efficiency and performance.

G\"odel's ontological proof has been analysed for the first-time with an unprecedent degree of detail and formality with the help of higher-order theorem provers. The following has been done (and in this order): A detailed natural deduction proof. A formalization of the axioms, definitions and theorems in the TPTP THF syntax. Automatic verification of the consistency of the axioms and definitions with Nitpick. Automatic demonstration of the theorems with the provers LEO-II and Satallax. A step-by-step formalization using the Coq proof assistant. A formalization using the Isabelle proof assistant, where the theorems (and some additional lemmata) have been automated with Sledgehammer and Metis.

Beachgoers in Australia can be a little less panicky about shark attacks this summer, because artificial intelligence-equipped drones will be monitoring the water for big scary fish. Developed by researchers at University of Technology Sydney, the AI system--dubbed Sharkspotter--can identify sharks and notify beachgoers when they're nearby. An Australian drone company called Westpac teamed up with the school to outfit its battery-powered Little Ripper Lifesaver unmanned helicopters with the technology in an effort to, hopefully, offer swimmers and surfers better protection. The AI-equipped drones will patrol "many main beaches in Australia" this summer, the school said in a news release. "The system will give an overhead warning to swimmers/surfers when a shark or a potential risk is detected, using an on-board megaphone attached to the drones," Professor Michael Blumenstein, head of the UTS School of Software, said in a statement, adding that the system "will create a positive impact for the public, making beach recreation much safer." Sharkspotter uses "cutting-edge deep neural networks and image processing techniques" to examine live video feeds in real time and detect the presence of sharks, and distinguish them from other marine life and objects, the school said.

Most marketers are confident the rise of artificial intelligence (AI) will not impact creative functions, according to research by The Drum, in association with social media analytics platform Sysomos. While over half of the marketers surveyed (61%) believe the integration of AI will result in a loss of jobs, almost two-thirds (63%) feel confident creative jobs will prove to be resistant to the threat of automation. But when asked about specific business areas marketers do not want AI to handle, creative briefs (43%) and recruitment (37%) came out on top. The findings come from The Drum Market Insight Report – Artificial Intelligence Edition. The Drum partnered with Sysomos to discover how marketers view AI and how it will impact their marketing agenda over the next five years. Over 200 marketers were surveyed to gauge opinions on AI and the automation of jobs, specific areas marketers do not want AI to handle, and investment in AI technology.

Artificial intelligence could predict Alzheimer's disease in a patient years before a doctor does, according to a new study by McGill University in Montreal. Researchers from the Douglas Mental Health University Institute's Translational Neuroimaging Laboratory at McGill were able to predict dementia by using AI techniques and big data to build an algorithm that can recognize signs of dementia two years in advance. Researchers used a single amyloid PET scan of the brain of patients at risk of developing Alzheimer's disease for the study. The early prognosis would give patients and families a chance to plan and manage treatment and care of the disease, researchers said. "By using this tool, clinical trials could focus only on individuals with a higher likelihood of progressing to dementia within the time frame of the study," Dr. Serge Gauthier, co-lead author of the study, said in a statement.

Almost every word has more than one meaning. Modern search engines solve this problem using knowledge bases. Yago was one of the first knowledge bases, developed by scientists at the Max Planck Institute for Informatics in Saarbrücken and the Télécom ParisTech in Paris. Last week, the researchers received an award for their work on Yago from the most important scientific journal in the field of artificial intelligence. Today, they are releasing Yago's source code.

Even if you can't make the live session, all registrants will have access after to the webinar, slides and a replay.

As we can see, the statsmodels library allows us to generate highly detailed output on a level similar to R, with additional statistics such as skew, kurtosis, R-Squared and AIC. While these readings can be generated through scipy or sklearn, doing so is a more intensive process and in many cases these statistics must be calculated individually.

I have implemented backprop before, but never quite understood it. After much hard work I finally found out how to derive the backprop equations, in a way that I could understand. Wanted to share my approach.