 |
August 30, 2017, 10:15:45 AM |
|
NeuroSeed is a decentralized platform that will provide proven data for Machine Learning. The project will solve the problem of trust between data providers and ML developers. The platform will combine functions (ready-made solutions), cloud and resources for providing computing power, data owners and neural network providers who are ready to give certain branches of their networks for faster creation of new products.
Mathematically proved that creating your own neural network does not require the necessity to use certain data sets to train the system every time. You can simplify the process by "pulling out" the needful data from existing networks previously created on the platform.
Ex., most chatbots are built on the same data and the differences are only in purpose or localization. We can take the already created chatbot with its project data and supplement them with our own data. In this case, we optimize the process and we'll be completely sure of the correctness of the data.
Recently, scientists have tested a program that uses Machine Learning to diagnose Alzheimer's disease using examinations of 149 test subjects. The algorithm (Cascaded Multi-view Canonical Correlation (CaMCCo)) processed various analyzes and surveys of the body, then assessed the parameters in two stages. First, he selects data that best separates sick people from healthy ones. And then it choose those variables that indicate the disease, separating people with moderate impairment from Alzheimer's disease itself.
Similar examinations and tests are done to determine the carcinoma and some more serious diseases. With the NeuroSeed platform, we could use the same data in several sphere of medicine.
|
