-
Machine Learning Technology Could Also Be Created From Human Neurons
The researchers at Bistable Recurrent Cell, University of Liege have found a monumental enhancement of Machine Learning technology using the different modes of operations of human neurons to make an artificial neuron with extraordinary proficiencies.
The researchers have surpassed the conventional ML processes to allow recurrent networks to study temporal relationships exceeding one thousand discrete time units, with other methods failing after a hundred. The study is also published in the PLOS One journal.
The ML processes have evolved to a massive scale due to AI; they offer a wide array of applications that we frequently engross in our everyday life. For instance, the use of the time series data, which has time as a present component of weather patterns, stock prices etc.
Time series is a part of ML that employs knowledge of past activities to predict future activities using a specific type of artificial neural network – recurrent neural network (RNN). This network facilitates storing information over a period of time for the time series to process it effectively. It also efficiently updates the memory as new data is received. Although, training these networks is extremely challenging as the memory capabilities are limited in time.
Nicola Vecoven, a doctoral student in the Systems and Modeling lab at the University of Liège and first author of the study, said that “We can imagine the example of a network that receives new information every day, but after the fiftieth day, we notice that the information from the first day had already been forgotten. However, human neurons are capable of retaining information over an almost infinite period of time thanks to the bi-stability mechanism. This allows neurons to stabilize in two different states, depending on the history of the electrical currents they have been subjected to, and do this for an infinite period of time. In other words, thanks to this mechanism, human neurons can retain a bit (a binary value) of information for an infinite time.”
The researchers were able to manufacture a new artificial neuron with an identical mechanism by employing the bi-stability mechanism and merging it with the recurrent artificial networks. The results were exceptional. They accomplished the learning of temporal relationships for about 1000 time steps with the team working to advance the memories of RNN even more.
Source: https://www.innovationnewsnetwork.com/creating-machine-learning-technology-from-human-neurons/12426/
-
0 Replies
Sorry, there were no replies found.
Log in to reply.