Fetching the paper…

The global optimum of shallow neural network is attained by ridgelet transform · Around