Second day of deep learning and tensorfow

Presence of decay factor

 

Kinear Regression continuity

                                                               Logistic Regression Two points

                                                               Classification piecewise

actual combat

numpy database

def compute-error-for-line-given-points(b,w,points):
    total-error = 0
    for i in range(0,len(point)):
       x= points[i,0]
       y=points[i,1]
       total-error += (y-( w*x+b)) **2
return total -error / float(len(points))

2. Handwritten numeral analysis  

one-hot problem

Step by step dimension reduction