Scientist creates an algorithm that is open-source for selecting the dictionary of a Neurointerface
Alexandra Bernadotte, Ph.D. and Associate Professor of Information Technologies and Computer Sciences, MISIS University has developed algorithms to increase the accuracy of robotic devices in recognizing mental commands. This is achieved by optimizing a dictionary. Robotic devices can transmit information using algorithms. The results were published in peer-reviewed scientific journal Mathematics.
Developers of systems that aim to improve human life face the challenge of improving object classification accuracy (audio, video, or electromagnetic signals), when compiling \”dictionaries\” for devices.
A voice assistant is the simplest of all. Audio and video transmission devices used for remote control an object within the line of sight zone only use a small set of commands. It is also important that the neural network-based commands classifier accurately understands the commands in the device dictionary and doesn’t confuse them. This also means that in the presence extraneous sounds, the recognition accuracy shouldn’t fall below a specific value.