TOKYO -- From evaluating biopsies in a hospital to whipping up a meal in a kitchen, machine learning is opening a new frontier for Japan's leading information technology companies.
Computers with machine learning systems can collect and analyze huge sets of data, then present that information to help people make better decisions.
IBM Japan has designed a system for the kitchen by suggesting recipes based on ingredients on hand as well as individual preferences and even allergies.
The system can turn voice commands into text on a screen, record information from grocery receipts, and formulate menus from a database of 12,000 recipes.
The company hopes to have a practical version of the system ready for the home by 2015, and sees it being incorporated into refrigerators, microwave ovens and other appliances. To this end, it will seek collaborations with appliance and household equipment manufacturers.
NEC has used machine learning for a system that can help doctors diagnose cancer. Shown a collection of more than 100,000 pictures of cancer cells and related diagnoses, the system compares these with new biopsies to spot suspected cancers and help pathologists make more accurate identifications. Given the shortage of pathologists in Japan, this could prove to be a tremendous aid.
Hitachi uses machine learning for its EMIEW2 service robot to help people find lost items, such as a watch misplaced somewhere in a room. With cameras on the ceiling and an Internet connection, the robot is always on alert. Having been taught ahead of time to recognize people's voices, remember various items and the layout of a room, it can quickly provide answers to such questions as "where's my watch?" Hitachi envisions the EMIEW2 being used in nursing homes and other places.