If you've been on the internet recently, you must have used or interacted with a deep learning program. One of the more popular ones is "ChatGPT-3" which was released by OpenAI.

In this article, we will be introducing this concept known as "deep learning".

What is deep learning ?

Deep learning is a branch of machine learning that runs on artificial neural networks.
These neural networks mimic the brain and learning patterns associated with it.

The deep learning algorithms are inspired by the structure and function of the brain thereby mitigating explicit programming as it learns to use data it receives over time, much like the human brain.

History

Deep learning has been around for a while. Neural networks were a major area of research in both neuroscience and computer science until 1969. It didn’t thrive as it is now because there wasn’t a lot of sufficient data and processing power for this concept to thrive as we have nowadays.

This tech enjoyed a resurgence in the 1980s but fell into eclipse again in the first decade of the century and has returned once again in what seems like a hype due to the rise in popularity of ground-breaking AI tools like DALL-E and ChatGPT, which are both products by Open AI.

In the past ten years, the best-performing artificial intelligence systems, such as Amazon's "Alexa", Microsoft's "Cortana", Apple's "Siri" and Google's automatic translator, have all resulted from deep learning.

How do neural networks work ?

Neural networks are a means of doing machine learning, in which a computer learns to perform tasks by analyzing past examples or data. Usually, data that is fed into these computers are hand labeled in advance for the computer to use these labels to identify similar patterns.

An example of this is the image-recognition system which is fed with thousands or even millions of images of cars, cats, houses, and so on, depending on the goal to be accomplished with this system. The system will find visual patterns in the image that consistently correlate with specific labels.

How Deep Learning Models Actually Work

Deep learning models have the ability to learn from data continuously, and this makes them well-suited for tasks such as navigation, speech recognition, Natural Language Processing, image recognition, and more.

To put things in perspective, deep learning is a subdomain of machine learning with accelerated computational power and very large data sets, and with this, deep learning algorithms can self-learn hidden patterns within data to make predictions.

The underlying architecture for deep learning was inspired by the structure of the brain. For this reason, quite a few fundamental terminologies within deep learning can be mapped back to neurology. Similar to how neurons form the fundamental building blocks of the brain, deep learning architecture also contains a computational unit that allows the modeling of non-linear functions called perceptron.

A neural network consists of thousands or even millions of simple processing nodes that are densely interconnected. The majority of today's neural networks are organized into layers of nodes and are "feed-forward", meaning that data moves through them in only one direction. An individual node might be connected to several nodes in the layer beneath it, from which it receives data, and several nodes in the layer above it to which it sends data.

To each of its respective incoming connections, a node will assign a "weight". When the network is active, the node receives a different data item i.e., a different number over each of its connections, and multiplies it by the associated weight. It then adds the resulting products together, yielding a single number. If that number is below a particular threshold value, it passes no data to the next layer, and if the number exceeds the threshold value, the node "fires", which in today's neural networks means sending the sum of the weighted inputs along all its outgoing connections.

While a deep learning model is being trained, all its weights and thresholds are initially set to random data to compensate for the unavailability of large data sets at the onset of the training.

The training data is consistently fed from the input layer, passing through the hidden layers, which process most of the data getting multiplied and added together in complex ways to yield unique or highly accurate results according to each input. This process is repeated until the model consistently recognizes data with similar labels and, as such, yields similar outputs.

In its most basic form, a neural network contains three layers: an input layer, a hidden layer, and an output layer.

Applications of Deep Learning

Deep learning as a technology is limitless and can be applied in almost any field. Some of these can include:

1. Image recognition: it can be used to train models to recognize shapes, images, and objects as well as to understand content and context. This is already being implemented in Gaming, retail, and tourism.

2. Advanced Map Navigation: Deep learning can be used to train advanced navigation systems. Some of this can be seen in Tesla cars and google's navigation systems.

3. Automatic text generation: A collection of text can be fed into a deep learning model, word-by-word, or character-by-character. Sequentially, the model is capable of learning to spell, write sentences, punctuate, or even capture a particular style of writing.

4. Natural Language Processing: Deep learning models are used to understand, generate and translate human languages. This can be done by feeding large amounts of vocal data sets to the model for training, repeating the process until the outputs become accurate and similar.

5. Robotics: Deep learning models are used to train robots and improve their ability to perceive and interact with the environment.

6. Finance: Deep learning models are used to analyze past data and make accurate predictions and also detect fraud.

Conclusion

To wrap it up, deep learning is extremely powerful and can be used to achieve great things. There are still plenty of theoretical and moral questions to be answered but armed with the information in this article, you are now ready to go deeper into your deep learning journey. Stay tuned for more articles in this series. Build "Deep" things.