Drones better than human rescuers at following mountain pathways

Curated on Posted on by Stefaan Verhulst

Springwise: “Every year in Switzerland, emergency centers respond to around 1,000 call outs for lost and injured hikers. It can often take hours and significant manpower to locate lost mountaineers, but new software for quadcopter drones is making the hunt quicker and easier, and has the potential to help find human survivors in disaster zones around the world.

The drone uses a computer algorithm called a Deep Neural Network. The program was developed by researchers at the University of Zurich and the Dalle Molle Institute for Artificial Intelligence. The drone uses the algorithm to learn trails and paths through a pair of small cameras, interpreting the images and recognizing man-made pathways. Even when working on a previously unseen trail, it was able to guess the correct direction in 85 percent of the cases. The drones’ speed and accuracy make them more effective than human trackers.

Drohnen-Scaramuzza-2
The researchers hope that eventually multiple small drones could be combined with human search and rescue missions, to cover more terrain and find people faster. The drones can cover terrain quickly and check hazardous areas to minimize risk to human workers, and its AI can identify paths and avoid crashing without any human involvement….(More)”

Forecasting Domestic Violence: A Machine Learning Approach to Help Inform Arraignment Decisions

Curated on Posted on by Stefaan Verhulst

Richard A. Berk, Susan B. Sorenson and Geoffrey Barnes in the The Journal of Empirical Legal Studies: “Arguably the most important decision at an arraignment is whether to release an offender until the date of his or her next scheduled court appearance. Under the Bail Reform Act of 1984, threats to public safety can be a key factor in that decision. Implicitly, a forecast of “future dangerousness” is required. In this article, we consider in particular whether usefully accurate forecasts of domestic violence can be obtained. We apply machine learning to data on over 28,000 arraignment cases from a major metropolitan area in which an offender faces domestic violence charges. One of three possible post-arraignment outcomes is forecasted within two years: (1) a domestic violence arrest associated with a physical injury, (2) a domestic violence arrest not associated with a physical injury, and (3) no arrests for domestic violence. We incorporate asymmetric costs for different kinds of forecasting errors so that very strong statistical evidence is required before an offender is forecasted to be a good risk. When an out-of-sample forecast of no post-arraignment domestic violence arrests within two years is made, it is correct about 90 percent of the time. Under current practice within the jurisdiction studied, approximately 20 percent of those released after an arraignment for domestic violence are arrested within two years for a new domestic violence offense. If magistrates used the methods we have developed and released only offenders forecasted not to be arrested for domestic violence within two years after an arraignment, as few as 10 percent might be arrested. The failure rate could be cut nearly in half. Over a typical 24-month period in the jurisdiction studied, well over 2,000 post-arraignment arrests for domestic violence perhaps could be averted….(More)”

Digital Decisions: Policy Tools in Automated Decision-Making

Curated on Posted on by Stefaan Verhulst

Ali Lange at CDT: “Digital technology has empowered new voices, made the world more accessible, and increased the speed of almost every decision we make as businesses, communities, and individuals. Much of this convenience is powered by lines of code that rapidly execute instructions based on rules set by programmers (or, in the case of machine learning, generated from statistical correlations in massive datasets)—otherwise known as algorithms. The technology that drives our automated world is sophisticated and obscure, making it difficult to determine how the decisions made by automated systems might fairly or unfairly, positively or negatively, impact individuals. It is also harder to identify where bias may inadvertently arise. Algorithmically driven outcomes are influenced, but not exclusively determined, by technical and legal limitations. The landscape of algorithmic decision-making is also shaped by policy choices in technology companies and by government agencies. Some automated systems create positive outcomes for individuals, and some threaten a fair society. By looking at a few case studies and drawing out the prevailing policy principle, we can draw conclusions about how to critically approach the existing web of automated decision-making. Before considering these specific examples, we will present a summary of the policy debate around data-driven decisions to give context to the examples raised. Then we will analyze three case studies from diverse industries to determine what policy interventions might be applied more broadly to encourage positive outcomes and prevent the risk of discrimination….(More)”

Political Speech Generation

Curated on Posted on by Stefaan Verhulst

Valentin Kassarnig at arXiv: “In this report we present a system that can generate political speeches for a desired political party. Furthermore, the system allows to specify whether a speech should hold a supportive or opposing opinion. The system relies on a combination of several state-of-the-art NLP methods which are discussed in this report. These include n-grams, Justeson & Katz POS tag filter, recurrent neural networks, and latent Dirichlet allocation. Sequences of words are generated based on probabilities obtained from two underlying models: A language model takes care of the grammatical correctness while a topic model aims for textual consistency. Both models were trained on the Convote dataset which contains transcripts from US congressional floor debates. Furthermore, we present a manual and an automated approach to evaluate the quality of generated speeches. In an experimental evaluation generated speeches have shown very high quality in terms of grammatical correctness and sentence transitions….(More)”

Yahoo Releases the Largest-ever Machine Learning Dataset for Researchers

Curated on Posted on by Stefaan Verhulst

Suju Rajan at Yahoo Labs: “Data is the lifeblood of research in machine learning. However, access to truly large-scale datasets is a privilege that has been traditionally reserved for machine learning researchers and data scientists working at large companies – and out of reach for most academic researchers.

Research scientists at Yahoo Labs have long enjoyed working on large-scale machine learning problems inspired by consumer-facing products. This has enabled us to advance the thinking in areas such as search ranking, computational advertising, information retrieval, and core machine learning. A key aspect of interest to the external research community has been the application of new algorithms and methodologies to production traffic and to large-scale datasets gathered from real products.

Today, we are proud to announce the public release of the largest-ever machine learning dataset to the research community. The dataset stands at a massive ~110B events (13.5TB uncompressed) of anonymized user-news item interaction data, collected by recording the user-news item interactions of about 20M users from February 2015 to May 2015.

The Yahoo News Feed dataset is a collection based on a sample of anonymized user interactions on the news feeds of several Yahoo properties, including the Yahoo homepage, Yahoo News, Yahoo Sports, Yahoo Finance, Yahoo Movies, and Yahoo Real Estate.

Our goals are to promote independent research in the fields of large-scale machine learning and recommender systems, and to help level the playing field between industrial and academic research. The dataset is available as part of the Yahoo Labs Webscope data-sharing program, which is a reference library of scientifically-useful datasets comprising anonymized user data for non-commercial use.

In addition to the interaction data, we are providing categorized demographic information (age range, gender, and generalized geographic data) for a subset of the anonymized users. On the item side, we are releasing the title, summary, and key-phrases of the pertinent news article. The interaction data is timestamped with the relevant local time and also contains partial information about the device on which the user accessed the news feeds, which allows for interesting work in contextual recommendation and temporal data mining….(More)”

Can crowdsourcing decipher the roots of armed conflict?

Curated on Posted on by Stefaan Verhulst

Stephanie Kanowitz at GCN: “Researchers at Pennsylvania State University and the University of Texas at Dallas are proving that there’s accuracy, not just safety, in numbers. The Correlates of War project, a long-standing effort that studies the history of warfare, is now experimenting with crowdsourcing as a way to more quickly and inexpensively create a global conflict database that could help explain when and why countries go to war.

The goal is to facilitate the collection, dissemination and use of reliable data in international relations, but a byproduct has emerged: the development of technology that uses machine learning and natural language processing to efficiently, cost-effectively and accurately create databases from news articles that detail militarized interstate disputes.

The project is in its fifth iteration, having released the fourth set of Militarized Dispute (MID) Data in 2014. To create those earlier versions, researchers paid subject-matter experts such as political scientists to read and hand code newswire articles about disputes, identifying features of possible militarized incidents. Now, however, they’re soliciting help from anyone and everyone — and finding the results are much the same as what the experts produced, except the results come in faster and with significantly less expense.

As news articles come across the wire, the researchers pull them and formulate questions about them that help evaluate the military events. Next, the articles and questions are loaded onto the Amazon Mechanical Turk, a marketplace for crowdsourcing. The project assigns articles to readers, who typically spend about 10 minutes reading an article and responding to the questions. The readers submit the answers to the project researchers, who review them. The project assigns the same article to multiple workers and uses computer algorithms to combine the data into one annotation.

A systematic comparison of the crowdsourced responses with those of trained subject-matter experts showed that the crowdsourced work was accurate for 68 percent of the news reports coded. More important, the aggregation of answers for each article showed that common answers from multiple readers strongly correlated with correct coding. This allowed researchers to easily flag the articles that required deeper expert involvement and process the majority of the news items in near-real time and at limited cost….(more)”

Predictive Analytics

Curated on Posted on by Stefaan Verhulst

Revised book by Eric Siegel: “Prediction is powered by the world’s most potent, flourishing unnatural resource: data. Accumulated in large part as the by-product of routine tasks, data is the unsalted, flavorless residue deposited en masse as organizations churn away. Surprise! This heap of refuse is a gold mine. Big data embodies an extraordinary wealth of experience from which to learn.

Predictive analytics unleashes the power of data. With this technology, the computer literally learns from data how to predict the future behavior of individuals. Perfect prediction is not possible, but putting odds on the future drives millions of decisions more effectively, determining whom to call, mail, investigate, incarcerate, set up on a date, or medicate.

In this lucid, captivating introduction — now in its Revised and Updated edition — former Columbia University professor and Predictive Analytics World founder Eric Siegel reveals the power and perils of prediction:

    • What type of mortgage risk Chase Bank predicted before the recession.
    • Predicting which people will drop out of school, cancel a subscription, or get divorced before they even know it themselves.
    • Why early retirement predicts a shorter life expectancy and vegetarians miss fewer flights.
    • Five reasons why organizations predict death — including one health insurance company.
    • How U.S. Bank and Obama for America calculated — and Hillary for America 2016 plans to calculate — the way to most strongly persuade each individual.
    • Why the NSA wants all your data: machine learning supercomputers to fight terrorism.
    • How IBM’s Watson computer used predictive modeling to answer questions and beat the human champs on TV’s Jeopardy!
    • How companies ascertain untold, private truths — how Target figures out you’re pregnant and Hewlett-Packard deduces you’re about to quit your job.
    • How judges and parole boards rely on crime-predicting computers to decide how long convicts remain in prison.
    • 183 examples from Airbnb, the BBC, Citibank, ConEd, Facebook, Ford, Google, the IRS, LinkedIn, Match.com, MTV, Netflix, PayPal, Pfizer, Spotify, Uber, UPS, Wikipedia, and more….(More)”

 

Daedalus Issue on “The Internet”

Curated on Posted on by Stefaan Verhulst

Press release: “Thirty years ago, the Internet was a network that primarily delivered email among academic and government employees. Today, it is rapidly evolving into a control system for our physical environment through the Internet of Things, as mobile and wearable technology more tightly integrate the Internet into our everyday lives.

How will the future Internet be shaped by the design choices that we are making today? Could the Internet evolve into a fundamentally different platform than the one to which we have grown accustomed? As an alternative to big data, what would it mean to make ubiquitously collected data safely available to individuals as small data? How could we attain both security and privacy in the face of trends that seem to offer neither? And what role do public institutions, such as libraries, have in an environment that becomes more privatized by the day?

These are some of the questions addressed in the Winter 2016 issue of Daedalus on “The Internet.”  As guest editors David D. Clark (Senior Research Scientist at the MIT Computer Science and Artificial Intelligence Laboratory) and Yochai Benkler (Berkman Professor of Entrepreneurial Legal Studies at Harvard Law School and Faculty Co-Director of the Berkman Center for Internet and Society at Harvard University) have observed, the Internet “has become increasingly privately owned, commercial, productive, creative, and dangerous.”

Some of the themes explored in the issue include:

  • The conflicts that emerge among governments, corporate stakeholders, and Internet users through choices that are made in the design of the Internet
  • The challenges—including those of privacy and security—that materialize in the evolution from fixed terminals to ubiquitous computing
  • The role of public institutions in shaping the Internet’s privately owned open spaces
  • The ownership and security of data used for automatic control of connected devices, and
  • Consumer demand for “free” services—developed and supported through the sale of user data to advertisers….

Essays in the Winter 2016 issue of Daedalus include:

  • The Contingent Internet by David D. Clark (MIT)
  • Degrees of Freedom, Dimensions of Power by Yochai Benkler (Harvard Law School)
  • Edge Networks and Devices for the Internet of Things by Peter T. Kirstein (University College London)
  • Reassembling Our Digital Selves by Deborah Estrin (Cornell Tech and Weill Cornell Medical College) and Ari Juels (Cornell Tech)
  • Choices: Privacy and Surveillance in a Once and Future Internet by Susan Landau (Worcester Polytechnic Institute)
  • As Pirates Become CEOs: The Closing of the Open Internet by Zeynep Tufekci (University of North Carolina at Chapel Hill)
  • Design Choices for Libraries in the Digital-Plus Era by John Palfrey (Phillips Academy)…(More)

See also: Introduction

Big Data Analysis: New Algorithms for a New Society

Curated on Posted on by Stefaan Verhulst

Book edited by Nathalie Japkowicz and Jerzy Stefanowski: “This edited volume is devoted to Big Data Analysis from a Machine Learning standpoint as presented by some of the most eminent researchers in this area.

It demonstrates that Big Data Analysis opens up new research problems which were either never considered before, or were only considered within a limited range. In addition to providing methodological discussions on the principles of mining Big Data and the difference between traditional statistical data analysis and newer computing frameworks, this book presents recently developed algorithms affecting such areas as business, financial forecasting, human mobility, the Internet of Things, information networks, bioinformatics, medical systems and life science. It explores, through a number of specific examples, how the study of Big Data Analysis has evolved and how it has started and will most likely continue to affect society. While the benefits brought upon by Big Data Analysis are underlined, the book also discusses some of the warnings that have been issued concerning the potential dangers of Big Data Analysis along with its pitfalls and challenges….(More)”

OpenAI won’t benefit humanity without data-sharing

Curated on Posted on by Stefaan Verhulst

 at the Guardian: “There is a common misconception about what drives the digital-intelligence revolution. People seem to have the idea that artificial intelligence researchers are directly programming an intelligence; telling it what to do and how to react. There is also the belief that when we interact with this intelligence we are processed by an “algorithm” – one that is subject to the whims of the designer and encodes his or her prejudices.

OpenAI, a new non-profit artificial intelligence company that was founded on Friday, wants to develop digital intelligence that will benefit humanity. By sharing its sentient algorithms with all, the venture, backed by a host of Silicon Valley billionaires, including Elon Musk and Peter Thiel, wants to avoid theexistential risks associated with the technology.

OpenAI’s launch announcement was timed to coincide with this year’s Neural Information Processing Systems conference: the main academic outlet for scientific advances in machine learning, which I chaired. Machine learning is the technology that underpins the new generation of AI breakthroughs.

One of OpenAI’s main ideas is to collaborate openly, publishing code and papers. This is admirable and the wider community is already excited by what the company could achieve.

OpenAI is not the first company to target digital intelligence, and certainly not the first to publish code and papers. Both Facebook and Google have already shared code. They were also present at the same conference. All three companies hosted parties with open bars, aiming to entice the latest and brightest minds.

However, the way machine learning works means that making algorithms available isn’t necessarily as useful as one might think. A machine- learning algorithm is subtly different from popular perception.

Just as in baking we don’t have control over how the cake will emerge from the oven, in machine learning we don’t control every decision that the computer will make. In machine learning the quality of the ingredients, the quality of the data provided, has a massive impact on the intelligence that is produced.

For intelligent decision-making the recipe needs to be carefully applied to the data: this is the process we refer to as learning. The result is the combination of our data and the recipe. We need both to make predictions.

By sharing their algorithms, Facebook and Google are merely sharing the recipe. Someone has to provide the eggs and flour and provide the baking facilities (which in Google and Facebook’s case are vast data-computation facilities, often located near hydroelectric power stations for cheaper electricity).

So even before they start, an open question for OpenAI is how will it ensure it has access to the data on the necessary scale to make progress?…(More)”