San Francisco

---

The Brain activity was fed into a computer system that is recognised, repeating features in the data

scientists will have to decode a Step Forward in their ability to what a person is saying just by looking at your brain waves when you speak.

they are trained to algorithms, The Brain -patterns into sets, in real time, and with word-error-rate as low as 3%.

so Far, this so-called "brain-machine-interfaces" had limited success in the decoding of neuronal activity.

The study

The previous efforts in this area were only able to decode fragments of spoken words or a small percentage of The Words in certain phrases.

Machine-learning specialists Dr. Joseph Makin of the University of California, San Francisco (UCSF), USA, and colleagues attempted to increase the accuracy.

Four volunteers to read the sentences aloud, while the electrodes record your brain activity.

The activity of The Brain fed was created in a computer system, which is a representation of regularly occurring features in the data.

in Limited, language

These patterns are likely with repeated features of speech, such as vowels, consonants or commands to the parts of The Mouth .

Another part of The System decodes this representation, word-for-word-sentences.

But, the authors admit that the study has limitations. For example, the decoded language, was limited to 30-50 sets.

"Although we should, as the decoder to learn and use the laws of language, it remains to show, how many data would be required for the expansion of our small languages to a more General form of the English language," the researchers wrote in their Nature Neuroscience paper.

But they add that the decoder is to classify simple sentences based on their structure. You know this, because its performance was improved by adding sentences from the training set that were not used in the tests themselves.

The Scientists say this proves that the-machine-interface is the identification of the individual words, not just sentences. To decrypt, In Principle , this means that it could be possible, sets of never met in a training set.

If you have been trained on the computer system, on the activity of The Brain and of The Speech of a person before and after the training to other volunteers, the decoding improves the results, suggesting that the technique will be transferable on humans.

follow Paul

brain, language, artificial intelligence

Source of news: bbc.com