Algorithm 1: DLSA Optimization Method for Neural Network Training
1	Initialize the network with a set of random parameters: weights, biases, Levenberg parameter µ
2	Compute Jacobian J and the approximated Hessian JTJ and the total sum of squared error
3	Update the weights and the biases Using the equation: xk + 1 = xk − [JTJ + μI]−1JTe
4	Recompute the total sum of squared error
5	Is the error performance Satisfactory?
6		if yes
7			Safe the training weights and biases
8		else
9			increase µ and re-calculate Δ θ and repeat the process from step 2 initialize the network until the last stage
10	end