This is basically the same setting as semantic segmentation so any architecture that solves this problem can be adapted for your application. You can have a look at papers like U-Net, or Fully Convolutional Neural Networks, whose implementations are available for Caffe (I would also be surprised if there weren't re-implementations in other frameworks like TensorFlow or Theano). On a side note, I am personally using the very naive approach that you are describing (extracting patches around each pixel) and I have found that the performance loss is not that critical for my application. It might still be a good idea for you to try it anyway before switching to more "exotic" architectures.

Yahoo's idea is for the billboard's ad content to be based on real-time information about a crowd of people, who could be commuters on a train platform. Hacked web giant Yahoo has filed a patent application for the ultimate ad-targeting system: a billboard that uses sensors to watch, listen and capture biometric data from the passing public. Yahoo, still in damage control from this week's claims that it helped the government spy on its email users, has filed a patent for smart technology that brings online ad-targeting capabilities to public billboards. The billboards would have cameras, microphones, motion-proximity sensors, and biometric sensors, such as fingerprint or retinal scanning, or facial recognition, according to the patent, which was filed last year but published on Thursday. The sensors would be used to measure engagement of passers-by. "For example, image data or motion-proximity sensor data may be processed to determine whether any members of the audience paused or slowed down near the advertising content, from which it may be inferred that the pause or slowing was in response to the advertising content (eg, a measurement of'dwell time')," Yahoo writes.

Up until very recently, computers needed a complicated and extremely precise set of instructions in order to accomplish even the simplest of tasks. Who among us remembers programming via punch cards? Or DOS? Computer programming languages have evolved over the years, but the biggest step has been moving towards the elimination of complicated programming. In other words, teaching computers to learn for themselves, dubbed machine learning.

In this post, I'll introduce you to the Julia programming language and a couple long-term projects of mine: Plots for easily building complex data visualizations, and JuliaML for machine learning and AI. Easily create strongly-typed custom data manipulators. "User recipes" and "type recipes" can be defined on custom types to enable them to be "plotted" just like anything else. We believe that Julia has the potential to change the way researchers approach science, enabling algorithm designers to truly "think outside the box" (because of the difficulty of implementing non-conventional approaches in other languages).

Some members of our team (including me) live by just 2 passions in life – Data Science & Movies! For us, slicing and dicing movies over Monday morning coffee is part of warming up ritual. So, we decided to do a poll among ourselves on the best movies related to data science and machine learning. We also thought that we would release the outcome of the results in form of an infographic. Needless to say there were heated debates and a few disappointed faces in our office! But I think the list we got to is a fairly representative of the thoughts of the group.

If you do not want to visit that page, you can return to the previous page.

October 7, 2016, 7:59 AM IBM's artificial intelligence technology, Watson, amazed the world five years ago when it won "Jeopardy." After that, Watson went on to help improve industries like education and transportation. Now Watson is discovering new insights for cancer patients, as IBM has enlisted 20 top cancer institutes to teach Watson.

Healthcare is a numbers game and the figures are running wild. Every indicator shows the creaking NHS cannot cope with rising demand and dwindling resources. But the relentless gloom over hospital waiting lists, budgetary shortfalls and demographic time bombs is being challenged by a fresh approach that could revolutionise personal and national health. A wave of innovation driven by artificial intelligence (AI) is being hailed as both a saviour of traditional healthcare and the dawn of a new era in the public's engagement with their own health. "Healthcare is one of the highest cost areas for all modern economies, which makes it ripe for AI as providers look for efficiency to care for patients," says Dan Housman, chief technology officer at ConvergeHEALTH by Deloitte. "Healthcare is complex as an industry and is generating vast volumes of data from imaging, genomics, sensors, daily care and scientific research.

If you do not want to visit that page, you can return to the previous page.