Three of the Coolest New Innovations in Machine Learning
As a devoted technophile, chances are good you are at least somewhat familiar with the concept of artificial intelligence and machine learning. Machine learning is an application of AI that allows systems to automatically learn and improve from experience, without being programmed. Lately, there have been a number of advancements in machine learning that are, to put it mildly, pretty darn cool. Check out the following examples of this innovative technology:
Machine Learning’s Positive Impact on Mobile Devices
Machine learning is now being used to enhance the technology that we use every single day. A perfect example of this is the Snapdragon mobile platforms from Qualcomm that enable on-device machine learning and also allow our mobile devices to rely less on the cloud. The Snapdragon platforms are engineered to adapt to everything you do on your smartphone with high-quality connections, real-time, on-device processing of info and more. For example, by using machine learning in the Snapdragon 835 Mobile Platform, you can expect superb performance from your smartphone without draining your battery. In addition, the mobile device’s camera can learn to recognize objects and detect surroundings, which makes for amazing-looking photos.
Face Recognition Software is Now Better than Ever
If you have a smartphone or other device that uses facial recognition to identify you, you might be tempted to put on a Batman mask or a big hat and sunglasses in order to fool this state of the art technology. Scientists at the University of Cambridge have trained a machine learning algorithm to locate 14 key facial points that will allow the face recognition software to correctly identify a person most of the time, no matter what they are doing to disguise their face. The researchers labeled 2,000 photos of people wearing hats, scarves, fake beards and other disguises to show the location of these key facial points. Thanks to machine learning, the algorithm looks at the subset of these images and learned how the disguised faces would correspond to a regular face. The system was then able to identify people in a scarf 77 percent of the time and in a cap, scarf and glasses over half of the time. While it’s not as perfect as recognizing undisguised faces, machine learning has definitely helped this technology to “see” through these disguise tricks.
Identifying Bird Songs Using Machine Learning
The Google Brain team used machine learning based software called Tensorflow in order to help researchers in New Zealand identify the calls of native birds. They used the Tensorflow model to label spectrograms and classify bird songs, all in real time; after going through 15,000 hours of audio that was recorded in and around Wellington, the machine learning-based system was able to identify the calls of the Kakariki and Hihi birds.
As artificial intelligence and machine learning become more advanced as well as mainstream, it will be interesting to see other ways that machine learning will start to help us in our everyday lives. From helping our mobile devices to be smarter than ever and learn our habits, recognizing our faces and becoming adept at recognizing bird songs from hours of audio, it is safe to say that machine learning will only become more amazing and useful as time goes on.