## Question

You work for a security firm that wants to use an artificial neural network to create a video facial recognition system. So you create a training set with hundreds of images of people that are found in your video footage. You initialize the artificial neural network with random weights assigned to all its connections. When you feed through the first few images the system does a terrible job identifying whether those people are included in the video. What would the artificial neural network now do to try and improve?

A. It will reinitialize and add random weights to all the connections.

B. It will adjust the weights of the connections to see if it does a better job making a prediction.

C. It will add weight to the data to do a better job identifying the image in the network.

D. It will add more layers to the output layer to see if it does a better job making a prediction.

## Answer

B. It will adjust the weights of the connections to see if it does a better job making a prediction.

## Explanation

The correct answer is **B. It will adjust the weights of the connections to see if it does a better job making a prediction**.

An artificial neural network is a system that consists of many interconnected nodes or neurons that process information. Each node has a set of inputs, weights, and a bias value. The inputs are the data that are fed into the node, the weights are the parameters that determine how much influence each input has on the output, and the bias is a constant value that shifts the output.

The output of a node is calculated by multiplying each input by its corresponding weight, adding them up, adding the bias, and then applying an activation function that determines whether the node fires or not. The activation function can be linear, nonlinear, or threshold-based.

The output of one node can be the input of another node in the next layer of the network. The network can have multiple hidden layers between the input layer and the output layer. The output layer produces the final prediction or classification of the network.

The goal of training an artificial neural network is to find the optimal values for the weights and biases that minimize the error between the network’s prediction and the desired output. This is done by using a learning algorithm such as backpropagation.

Backpropagation is a technique that calculates the gradient of the error with respect to each weight and bias in the network, and then updates them in the opposite direction of the gradient. This means that the weights and biases are adjusted slightly to reduce the error.

The process of backpropagation involves two steps: forward propagation and backward propagation. In forward propagation, the network takes an input data and produces an output prediction. The error is then computed by comparing the prediction with the desired output.

In backward propagation, the error is propagated back through the network, starting from the output layer and going back to the input layer. At each node, the partial derivative of the error with respect to each weight and bias is calculated using the chain rule of calculus. The partial derivatives indicate how much each weight and bias contributes to the error.

The weights and biases are then updated by subtracting a fraction of their partial derivatives from their current values. The fraction is called the learning rate and it controls how fast or slow the network learns.

By repeating this process for many input data, the network learns to adjust its weights and biases to minimize the error and improve its prediction accuracy.

Some key points to remember are:

- An artificial neural network consists of many nodes or neurons that process information using weights, biases, and activation functions.
- The goal of training an artificial neural network is to find the optimal values for the weights and biases that minimize the error between the network’s prediction and the desired output.
- Backpropagation is a technique that calculates the gradient of the error with respect to each weight and bias in the network, and then updates them in the opposite direction of the gradient.
- The weights and biases are adjusted slightly to reduce the error at each iteration of training.
- The learning rate controls how fast or slow the network learns.

## Reference

- Optimization the Initial Weights of Artificial Neural Networks via Genetic Algorithm Applied to Hip Bone Fracture Prediction (hindawi.com)
- backpropagation – Neural networks – adjusting weights – Data Science Stack Exchange
- Weight (Artificial Neural Network) Definition | DeepAI
- Introduction to neural networks — weights, biases and activation | by Andrea D’Agostino | MLearning.ai | Medium
- computational modeling – What do the weights of an artificial neural network represent in biological neurons? – Psychology & Neuroscience Stack Exchange
- Applying Artificial Neural Networks for Face Recognition (hindawi.com)
- An optimized solution for face recognition | MIT News | Massachusetts Institute of Technology
- Deep learning, neural networks, algorithms boost facial recognition accuracy | VentureBeat

The latest Generative AI Skills Initiative certificate program actual real practice exam question and answer (Q&A) dumps are available free, helpful to pass the Generative AI Skills Initiative certificate exam and earn Generative AI Skills Initiative certification.