# Neural Network: How it works and its industry use cases

# What are Neural networks?

**Neural networks** are a series of algorithms that mimic the operations of a human brain to recognize relationships between vast amounts of data. They are used in a variety of applications in financial services, from forecasting and marketing research to fraud detection and risk assessment.

# How a Neural Network Works?

A neural network has many layers. Each layer performs a specific function, and the complex the network is, the more the layers are. That’s why a neural network is also called a multi-layer perceptron.

The purest form of a neural network has three layers:

- The input layer
- The hidden layer
- The output layer

As the names suggest, each of these layers has a specific purpose. These layers are made up of nodes. There can be multiple hidden layers in a neural network according to the requirements. The input layer picks up the input signals and transfers them to the next layer. It gathers the data from the outside world.

The hidden layer performs all the back-end tasks of calculation. A network can even have zero hidden layers. However, a neural network has at least one hidden layer. The output layer transmits the final result of the hidden layer’s calculation.

Like other machine learning applications, you will have to train a neural network with some training data as well, before you provide it with a particular problem. But before we go more in-depth of how a neural network solves a problem, you should know about the working of perceptron layers first:

# How do Perceptron Layers Work?

A neural network is made up of many perceptron layers; that’s why it has the name ‘multi-layer perceptron.’ These layers are also called hidden layers of dense layers. They are made up of many perceptron neutrons. They are the primary unit that works together to form a perceptron layer. These neurons receive information in the set of inputs. You combine these numerical inputs with a bias and a group of weights, which then produces a single output.

For computation, each neuron considers weights and bias. Then, the combination function uses the weight and the bias to give an output (modified input). It works through the following equation:

combination = bias +weights * inputs

After this, the activation function produces the output with the following equation:

output = activation(combination)

This function determines what kind of role the neural network performs. They form the layers of the network. The following are the prevalent activation functions:

## The Linear Function

In this function the output is only the combination of the neuron:

activation = combination

## The hyperbolic Tangent Function

It is the most popular activation function among neural networks. It is a sigmoid function, and it lies between -1 and +1:

activation = tanh(combination)

## The Logistic Function

The logistic function is quite similar to the hyperbolic tangent function because it is a kind of sigmoid function, as well. However, it is different because it lies between 0 and 1:

activation = 11 + e-combination

## The Rectified Linear Unit Function

Just like the hyperbolic tangent function, the rectified linear unit function is also prevalent. Another name for the rectified linear unit function is ReLU. ReLU is equal to the combination when it is equal to or greater than zero, and it’s negative if the combination is lower than (negative) zero.

# So, How Does a Neural Network Work Exactly?

Now that you know what is behind a neural network and how it works, we can focus on the working of a neural network.

Here’s how it works:

- Information is fed into the input layer which transfers it to the hidden layer
- The interconnections between the two layers assign weights to each input randomly
- A bias added to every input after weights are multiplied with them individually
- The weighted sum is transferred to the activation function
- The activation function determines which nodes it should fire for feature extraction
- The model applies an application function to the output layer to deliver the output
- Weights are adjusted, and the output is back-propagated to minimize error

The model uses a cost function to reduce the error rate. You will have to change the weights with different training models.

- The model compares the output with the original result
- It repeats the process to improve accuracy

The model adjusts the weights in every iteration to enhance the accuracy of the output.

# Use Cases of Neural Networks

Artificial **Neural Networks** can be **used** in a number of ways. They can classify information, cluster data, or predict outcomes. ANN’s can be **used** for a range of tasks. These include analyzing data, transcribing speech into text, powering facial recognition software, or predicting the weather.

## Top Companies using Artificial Neural Network(ANN)

**Nvidia Corp. (NVDA)****Alphabet (GOOG, GOOGL)****Salesforce.com (CRM)****Amazon.com (AMZN)****Microsoft Corp. (MSFT)****Twilio (TWLO)****IBM (IBM)****Facebook (FB)**

## Case Study: Netflix

Netflix Inc. isn’t going to leave the success of its series and films to chance — and analysts say its stock should be rewarded.

The company wants to be able to “combine great story telling and the great technological aspects,” Chief Executive Reed Hastings told MarketWatch in 2015. “That’s where we want to be.”

Netflix’s use of **convolutional neural network** and proprietary algorithms, which is essentially deep machine learning used to **analyze visual imagery**, is a prime example of its approach.

And it’s just that approach that grabbed the attention of Wells Fargo analysts Ken Sena and Marci Ryvicker. They initiated coverage of Netflix NFLX, +2.19%** **on Wednesday with an overweight rating and a $230 12-month price target, which is the highest price target among analysts covering the stock, according to FactSet.

Sena and Ryvicker said improvements and advancements in** neural networks **allow Netflix to effectively push recommended shows and movies to subscribers and even use data to make decisions in the production and acquisition of content.

Brian David Johnson, futurist in residence at Arizona State University, said people have been able to analyze video on a frame-by-frame and pixel-by-pixel basis for years — he even wrote a book about it in 2009. But there were roadblocks in content licensing that made innovation tough, until now.

“Netflix has been working solidly to come up with algorithms to match consumers with their content — also Netflix has a lot more power than they did back in 2009 to get people to allow them to search their video,” Johnson said in an email to MarketWatch. “The advances in [**artificial intelligence**] and** neural networks** means that they can now make sense of that data.

“Essentially, they are looking for patterns in the data that equal the right output they are looking for. The scope and scale of **AI** allows them to do this in an unprecedented fashion.”

In addition to using the collected data for platform improvements such as adding auto play, a “skip intro” button, customized trailers and changing its stars rating system to thumbs up or down, Ryvicker and Sena said Netflix could use the capability for decision-making concerning content supply.

Ryvicker and Sena wrote that through advancements in convolutional neural networks, Netflix can detect and analyze underlying scene elements that drive viewer engagement. That data-driven approach, they wrote, can inform what content Netflix licenses and provide insight into production.

“We see this as an important factor driving Netflix’s achievement around originals, with renewal rates roughly three times that of what the traditional TV networks have produced” Ryvicker and Sena wrote, also noting Netflix’s 91 Emmy nominations — second-most behind HBO. “In addition, Netflix itself estimates that these efficiencies, combined with higher subscriber retention, saves the company over $1 billion each year.”

Netflix is joining the likes of Facebook, Google and Microsoft in exploring deep learning algorithms to improve their user experience.

Netflix engineers Xavier Amatriain, Justin Basilico and Alex Chen explained the new endeavor in a blog post detailing their research into how deep learning could improve the Netflix movie and TV recommendation engine. Yet unlike larger companies such as Facebook or Google, Netflix is running its algorithms on Amazon Web Services, where it already hosts its streaming service, rather than building custom infrastructure.

Deep learning is a branch of artificial intelligence that aims to solve complex problems using computer systems that mimic the structure and behavior of the human brain. While Netflix hasn’t yet revealed how exactly they’ll apply the algorithms to recommendations, the engineers did break down how they’ll use GPUs to build the deep learning neural networks.

“In architecting our approach for leveraging computing power in the cloud, we sought to strike a balance that would make it fast and easy to train neural networks,” the engineers wrote. “We sought out to implement a large-scale neural network training system that leveraged both the advantages of GPUs and the AWS cloud. We have the capacity to use many GPU cores, CPU cores and AWS instances… In our solution, we take the approach of using GPU-based parallelism for training and using distributed computation for handling hyperparameter tuning and different configurations.”