Deep neural networks have seemingly endless applications.
One of the first deep learning networks developed were artificial neural networks (ANNs).
Artificial Neural Networks (ANNs)
Inspired by the human brain, the structure of ANNs was developed to try and mimic the way the human brain works.
They were first being proposed in 1943 by mathematicians Warren McCulloch and Walter Pitts. Since then,the big data revolution and increased processing power of computers, have meant that ANNs have been applied to give powerful insights.
Artificial Neural Networks are pretty awesome, to be honest.
Convolutional Neural Networks (CNNs)
Then came convolutional neural networks, inspired by the brain and the eye.
Convolutional Neural Networks (CNNs) are an area within of deep learning that have proved influential for image processing.
CNNs are used to convert images into a format that can be processed by a computer.
The CNN works by first mapping out the features of an image. Then the network then uses some mathematical tricks to convert this map into a feature vector that can then be processed.
Next, the feature vector is fed into other deep learning networks to understand what it is telling us.
Being able to convert images into data that can be understood and analyzed by machine learning opens this space up for many novel applications.
What have we learned about the world thanks to these neural network technologies?
Ok, so now we know a bit about how they work, it’s time to understand the applications of neural network technologies.
Ready to see some cool stuff?
Of course, you are!
From predicting the weather to understanding dementia, this technology has been used to do it all.
Below is a list of 27 ways that Neural Networks have been used that you might not have expected
Applications for Understanding Nature
The challenges of climate change are not to be underestimated.
One of the most exciting areas of research using neural networks is looking to better understand our planet. Here we have many uses of deep learning techniques to understand nature and the science of the world around us.
- Predicting the locations of oil flow rates – useful for oil miners
- Predicting the Weather – creating more accurate local predictions
- Modelling nature – understanding the environment so we can protect it
- Optimizing Air Ventilation air requirements for coal Mining – protecting people under ground
- Land classification – identifying useful features to help understand the land
- Environmental modeling to understand the impact of climate change
- Geotagging of photo
Neural Networks for Understanding Human Beings
As regular readers will know, I am passionate about the use of machine learning techniques for medicine. Leveraging the power of big data to understand how our bodies work allows us to make
- Predicting heart attacks to help prevent them
- Facial expression recognition – now you can finally know how they feel
- Medical research – there are many ways deep learning is helping make medical research and drug testing more efficient. It has been used to improve medial trial testing protocol.
- Lung cancer – Lung cancer kills, this process helps identify risk factors, types of cancer, and optimises treatments.
- Dementia detection – detecting dementia is tough and key to treatment options. Here deep neural networks can help
ANNs and CNNs Working in Business
Deep neural networks are helping to make businesses more efficient. Being able to
- Detecting Fraud – You can’t hide with deep learning around!
- Business optimisation and prediction models for marketing
- Forecasting demand to optimize supply chain for efficiency
- Urban modelling – understanding cities and their populations
- Modelling the rises and falls of the Stock market – Being the best at predicting can make you rich!
- Determining risk for Credit scoring and insurance. Understanding data trends to identify risk.
- Identifying corruption – going deep to detect signals indicating something funny is going on
- Making research more efficient – this is a general one to capture how neural networks and deep learning are making research processes more efficient. This is done by minimising the research required in narrowing down potential solutions.
New applications of Neural Networks
In addition to applications to understand the world, neural networks are also used in many new applications.
- Self-driving cars – using a variety of techniques including computer vision and reinforcement learning to train the car
- Text recognition – being able to pick out text from an image
- Google translate app – apps that are able to pick out text and translate use deep learning techniques
- Category detection – understanding what category something falls in. Hot dog or no hot dog?
- Facial recognition for surveillance and tracking – it’s not all fun and Facebook photos
- For cyber security -using neural networks to tackle crime in multiple ways.
Still some more to go!
- AR: All Augment Reality applications require some form of deep learning. For a computer to see it needs to be able to process impages. As a result it requires a CNN. This article covers different applications of augmented reality
- Passport control – another example of facial recognition in action. Hopefully one day they will have enough training data to make it efficient!
So there you have it. You must agree there are so many different and innovative applications of deep neural networks.
Which is your favourite?