Machine Learning: AI-Alerts


Why even a moth's brain is smarter than an AI

#artificialintelligence

One of the curious features of the deep neural networks behind machine learning is that they are surprisingly different from the neural networks in biological systems. While there are similarities, some critical machine-learning mechanisms have no analogue in the natural world, where learning seems ...


Missing data hinder replication of artificial intelligence studies

#artificialintelligence

The same algorithm can learn to walk in wildly different ways. Last year, computer scientists at the University of Montreal (U of M) in Canada were eager to show off a new speech recognition algorithm, and they wanted to compare it to a benchmark, an algorithm from a well-known scientist. The only ...


Guide - Machine Learning The F# Software Foundation

#artificialintelligence

F# is well-suited to machine learning because of its efficient execution, succinct style, data access capabilities and scalability. F# has been successfully used by some of the most advanced machine learning teams in the world, including several groups at Microsoft Research. This guide includes resources related to machine learning programming with F#. To contribute to this guide, log on to GitHub, edit this page and send a pull request. Note that the resources listed below are provided only for educational purposes related to the F# programming language. The F# Software Foundation does not endorse or recommend any commercial products, processes, or services. Therefore, mention of commercial products, processes, or services should not be construed as an endorsement or recommendation. Several F# machine learning packages are available. Some are accessed through F#'s interoperability mechanisms to R, Python and Java. .NET packages can be found by searching on nuget.org. For example: Accord.MachineLearning - Contains Support Vector Machines, Decision Trees, Naive Bayesian models, K-means, Gaussian Mixture models and general algorithms such as Ransac, Cross-validation and Grid-Search for machine-learning applications. This package is part of the Accord.NET Framework. See also First steps with Accord.NET SVM in F# R Packages - All R packages can be accessed through the RProvider for F#. See, for example, F# Neural Networks with the RProvider and Deedle Encog Machine Learning Framework - An advanced neural network and machine learning framework. Encog contains classes to create a wide variety of networks, as well as support classes to normalize and process data for these neural networks. Encog trains using multithreaded resilient propagation. Encog can also make use of a GPU to further speed processing time. A GUI based workbench is also provided to help model and train neural networks. See, for example, ENCOG Neural Network XOR example in F# Hype - An experimental deep learning library, where you can perform optimization on compositional machine learning systems of many components, even when such components themselves internally perform optimization. Underlying computations are run by a BLAS/LAPACK backend (OpenBLAS by default).


How Amazon Rebuilt Itself Around Artificial Intelligence

#artificialintelligence

In early 2014, Srikanth Thirumalai met with Amazon CEO Jeff Bezos. Thirumalai, a computer scientist who'd left IBM in 2005 to head Amazon's recommendations team, had come to propose a sweeping new plan for incorporating the latest advances in artificial intelligence into his division. He arrived ar...


How an A.I. 'Cat-and-Mouse Game' Generates Believable Fake Photos

#artificialintelligence

The woman in the photo seems familiar. She looks like Jennifer Aniston, the "Friends" actress, or Selena Gomez, the child star turned pop singer. But not exactly. She appears to be a celebrity, one of the beautiful people photographed outside a movie premiere or an awards show. And yet, you cannot...


Management AI: Bias, Criminal Recidivism, And The Promise Of Machine Learning

#artificialintelligence

Criminal recidivism is when a released criminal goes back to crime. From charging crimes through probation, the criminal justice system is constantly looking for ways to better predict which criminals are more likely to remain legal on release and who is a risk of recidivism. Bias can create inaccuracies through weighing variables incorrectly, and machine learning might provide a way of limiting bias and improving recidivism predictions. A recent study by Julia Dressel and Hany Farid, published in Science Advances, points to the limitations of deterministic algorithms with fixed parameters for the task of such predictions. The study analyzes the Correctional Offender Management Profiling for Alternative Sanctions (COMPAS) software, a package used by court systems to predict the likelihood of recidivism in criminal defendants.


Google's self-training AI turns coders into machine-learning masters

#artificialintelligence

Google just made it a lot easier to build your very own custom AI system. A new service, called Cloud AutoML, uses several machine-learning tricks to automatically build and train a deep-learning algorithm that can recognize things in images. The technology is limited for now, but it could be the start of something big. Building and optimizing a deep neural network algorithm normally requires a detailed understanding of the underlying math and code, as well as extensive practice tweaking the parameters of algorithms to get things just right. The difficulty of developing AI systems has created a race to recruit talent, and it means that only big companies with deep pockets can usually afford to build their own bespoke AI algorithms.


Google Sells A.I. for Building A.I. (Novices Welcome)

#artificialintelligence

Google has been using artificial intelligence to build other artificially intelligent systems for the last several months. Now the company plans to sell this kind of "automated machine learning" technology to other businesses across the globe. On Wednesday, Google introduced a cloud-computing service that it bills as a way to build a so-called computer vision system that suits your particular needs -- even if you have little or no experience with the concepts that drive it. If you are a radiologist, for example, you can use CT scans to automatically train a computer algorithm that identifies signs of lung cancer. If you run a real estate website, you can build an algorithm that distinguishes between living rooms and kitchens, bathrooms and bedrooms.



Can Enterra's Advanced AI Systems Stop The Fake News Epidemic?

Forbes - Business

The simplest way to eliminate the spread of fake news would be to limit ourselves to a small group of mainstream publishers who do all their own reporting and fact-checking. The counterargument, of course, is that an open and democratic society allows for a wide range of voices, not just the ones a small cabal of editors deem acceptable. Fake news promises to destroy this system and undermine trust and democracy, which is why addressing fake news has become one of the tech industry's most significant and important challenges. His initial focus, post-9/11, was on national security, which is how he first become intrigued by the advantages AI offers in analyzing complex data sets. As 2017's fake news scandals grew, DeAngelis was approached by leading media industry veteran Greg D'Alba, CEO of VIDL News, to apply the same type of analysis Enterra was using to control the complex value chains of Fortune 500 companies to the media industry, where D'Alba saw a growing need to verify and validate news stories.