. Neural Networks

from math import exp
from scipy.special import expit
import numpy as np
class ANN():
    def __init__(self) -> None:
        self.input_nodes=None
        self.weights_dict={}
        self.activation_dict={}
        self.outputs_dict={}

    def apply_activation(self,array,activation):
        if activation=='sigmoid':

            return 1/(1+np.exp(array))
        elif activation=='relu':
            np.maximum(0,array)
            
    def add_input_layer(self,nodes):
        self.input_nodes=nodes
    def add_layer(self,nodes,activation):
        if not self.weights_dict:
            self.weights_dict[1]=np.random.rand(0,1,[self.input_nodes,nodes])
            self.activation_dict[1]=activation
        else:
            self.weights_dict[max(self.weights_dict.keys())+1]=np.random.rand(0,1,[len(self.weights_dict[max(self.weights_dict.keys())]),nodes])
            self.activation_dict[max(self.activation_dict.keys())+1]=activation
    def fwd_pass(self,input_metrix,activation):
        """use 0 for input layer in layers vairable """
        for i in self.weights_dict.keys():
            if i==1:
                self.outputs_dict[i]=self.apply_activation(np.dot(input_metrix,self.weights_dict[i]),activation)
            else:
                self.outputs_dict[i]=self.apply_activation(np.dot(self.outputs_dict[i-1],self.weights_dict[i-1]),activation)
    def backwd_pass(self):
        pass

import numpy as np
np.random.uniform(0,1,[2,2])

inp=np.array([[1,2]])
weights=np.ones([2,3])
np.dot(inp,weights)

import numpy as np

a = np.array([1, 2, 3])
b = np.array([4, 5, 6])

# Compute the cross product
result = np.cross(a, b) 
# Output: array([-3,  6, -3])
result