Computer Science News Archive
An artificial intelligence system has correctly predicted the outcomes of hundreds of cases heard at the European Court of Human Rights, UCL researchers have claimed. The AI predicted the verdicts to an accuracy of 79%, according to the scientists involved.
AI is increasingly being used in fields such as journalism, law and accountancy, though critics said no AI would be able to understand the nuances of a legal case. The study, which was conducted by researchers at UCL Computer Science, led by Dr Nikolaos Aletras of the Media Futures Research Group, and the universities of Sheffield and Penn... [more]
Computer Science Research Student Jeremiah Onaolapo and colleagues from UCL's Information Security Research Group decided to find out how quickly criminals react once they get access to an online account.
The team set up 100 Gmail accounts and then accidentally-on-purpose shared their login credentials on forums and sites that data traders are known to frequent. The accounts were made to look "live" by having message threads, alerts and updates flow through them. They were also surreptitiously locked down to limit abuse.
Jeremiah Onaolapo was sure the webmail accounts would be tempting becaus... [more]
UCL Business spinout SmartTarget Ltd announces the launch of a new device for the detection of prostate cancer. The CE certified SmartTarget system is an image fusion device for guiding targeted, transperineal prostate biopsy.
Designed and developed by world leading urologists and biomedical engineers, the technology enables image-directed tumour targeting during minimally invasive biopsy procedures for prostate cancer patients and has undergone extensive clinical evaluation in over 200 patients at UCLH.
Users of the SmartTarget software benefit from an interface that is intuitive and easy-to-... [more]
Haitham Baomar and his colleague Peter Bentley are developing a special kind of autopilot: one that uses a “machine learning” system to cope when the going gets tough, rather than ceding control to the crew.
As reported by The Economist, today’s autopilots cannot be trained because they are “hard coded” programs in which a limited number of situations activate well-defined, pre-written coping strategies — to maintain a certain speed or altitude, say. A list of bullet points (which is what such programs amount to) does not handle novelty well: throw a situation at the computer that its programm... [more]
The Royal Academy of Engineering is the UK’s national academy for engineering and brings together the most successful and talented engineers from across the engineering sectors for a shared purpose: to advance and promote excellence in engineering.
The Academy's Fellowship represents the nation’s best engineering researchers, innovators, entrepreneurs, business and industry leaders. Election to the Academy is by invitation only; about 50 Fellows are elected each year by peer review from nominations made by existing Fellows. They are distinguished by the title "Fellow of the Royal Academy of ... [more]
UCL-CS Research Student Aron Monszpart has been awarded the Google PhD Fellowship in Machine Perception to continue exploring how computers can learn about their environment from observing how humans use it. Aron is a member of the Virtual Environments and Computer Graphics Research Group.
Aron's research topic is 3D scene understanding with unseen priors. We as humans use many other cues than just vision to interpret our environment, something that's still lacking in scene understanding performed by machines. Aron's research investigates how simple assumptions about the environment can yield... [more]
A team of Computer Science Undergraduates and researchers have won award from Facebook AI Research for work presented at the RepEval 2016 workshop. The project, led by Pontus Stenetorp, Research Associate, uses an enjoyable language-based game to collect linguistic annotations and also to introduce a novel way to evaluate AIs that communicate using natural language.
Congratulations to the student team comprised of Iuliana-Elena Parasca, Andreas Lukas Rauter, Jack Roper and Aleksandar Rusinov; and thanks to and Guillaume Bouchard and Sebastian Riedel for their input too.
The project was part of... [more]
Online databases now contain huge numbers of images. However there is only limited ability to search within these databases for an image with particular characteristics. Systems may rely on images being annotated with meaningful descriptions - "metadata".
Peter Wonka (of King Abdullah University of Science and Technology (KAUST) and Paul Guerrero and Niloy Mitra (both from University College London) wanted to add something more powerful to the currently limited repertoire of image search tools without adding extra metadata to existing images. They devised a query tool they call a relation-aug... [more]