Published on October 14th, 2016 | by Joshua New0
10 Bits: the Data News Hotlist
This week’s list of data news highlights covers October 8-14, 2016 and includes articles about a program in New Orleans that helps eliminate blighted buildings and a 3D printing system that uses AI to correct mistakes.
The White House has released two new reports detailing how the federal government should promote the development and adoption of artificial intelligence (AI). The reports detail recommendations and research and development priorities for the government that can help spur the growth of AI, such as promoting the development of data science skills and encouraging agencies to adopt AI to better serve the public.
The American Society of Clinical Oncology has launched a data sharing platform called CancerLinQ that allows oncologists to easily aggregate and share data from electronic health records and medical devices. To date, 68 oncology practices have signed on to use CancerLinQ, and they can use this data to better manage patient care. For example, CancerLinQ can make it easier for oncologists to ensure their patients are receiving appropriate treatments, regardless of where those treatments are administered.
The Iowa Department of Transportation has designated a portion of Interstate 380 (I-380) as a test corridor for autonomous vehicles. Beginning in 2017, the state will fully map a portion of I-380 to develop a testing platform for autonomous vehicles with help from the Chicago location technology company HERE. The state selected the stretch of I-380, between Iowa City and Cedar Rapids, because it has both rural and urban portions, as well as a high level of traffic, making it useful to comprehensively test the technology.
The city of New Orleans’ Office of Performance and Accountability (OPA) has been able to eliminate over 15,000 blighted properties from 2010 to 2015 as well as clear a 1,500 property backlog thanks to a data-driven project called BlightSTAT. When OPA receives a complaint about an abandoned or derelict building, it has to investigate the property and conduct a hearing against the owners if they do not bring the property up to code so the city can foreclose on or demolish the property, which can be laborious and time consuming. Through BlightSTAT, a machine learning algorithm analyzes property inspection and case data to predict which property owners will likely voluntarily bring their buildings up to code if the city were to simply send them a warning, rather than conduct a hearing, and which will require a hearing for the blighted building to be repaired or destroyed.
Google’s AI research arm DeepMind has developed a “differentiable neural computer,” a system that can perform the advanced analysis of a neural network as well as learn to identify and store certain information in its memory that it thinks could help solve future problems. The approach combines aspects from traditional computer systems, which are adept at reading and writing data to its memory, and neural networks, which are adept at learning from examples. With both of these abilities, a differentiable neural computer can learn how to remember solutions to problems and potentially apply them in different scenarios.
The U.S. Department of Justice has announced it will begin collecting national data in 2017 on uses of force by police, including police shootings, as there is little data available to the public or policymakers about these instances. The Department of Justice will require federal agents to report data about uses of force and work with local police departments to do the same, though local police departments are not legally required to report this data. To encourage participation, the Department of Justice will allocate $750,000 for the Police Data Initiative to help local departments collect this data and make it publicly available.
Mapping company Esri has partnered with navigation app Waze to make it easier for government agencies to share open geospatial data to improve transportation systems. Governments using Esri will be now be able to exchange data between the Esri platform and the Waze Connected Citizens Program, which allows Waze users and municipal governments to share traffic information with each other in real time. With easier access to real-time traffic data and high quality geospatial data, governments can make more informed decisions about transportation, such as better prioritizing road crews and emergency services.
IBM has announced that beginning next year, its U.S. employees with cancer will be able to use its Watson cognitive computing platform to identify clinical trials and treatments that are most likely to effectively treat their specific type of cancer. Watson will analyze participating employees’ health records and the sequencing data of their cancers and provide them and their doctors with information on drugs and clinical trials that could help.
London startup Ai Build has developed an industrial 3D printing system that uses AI to analyze and correct its mistakes. 3D printing complex objects can be time consuming, as rushing the process can result in mistakes that reduce a structure’s integrity. Ai Build relies on cameras attached to a 3D printer’s robotic arm and computer vision algorithms that monitor objects as its system prints them to quickly identify signs of a defect or mistake so it can take corrective action. Ai Build says that the ability to identify and correct errors allows its system to print objects much more quickly, cutting printing time in half.
Xerox has developed technology called the Vehicle Passenger Detection System that can detect the number of passengers in passing cars on the highway to determine if it is sufficient for them to be driving in a high-occupancy vehicle (HOV) lane and help police crack down on cheaters. The system uses cameras that take a picture of a passing vehicle and algorithms that can identify how many people are in the car—if there are too few, the system will flag the car’s license plate for police to send the car’s owner a fine. In a test on a highway in California, the system was able to identify drivers violating HOV rules with 95 percent accuracy, significantly better than human spotters, which are just 36 percent accurate.