Portal Sono Dicas Novidades da Tecnologia para Ti!
Deep learning is a neural network with three or more layers that attempt to simulate the behavior of the human brain. The technology allows the machine to “learn” in a deeper way than ordinary artificial intelligence (AI).
Deep learning models can achieve state-of-the-art accuracy, sometimes exceeding human-level performance. The technology is used in autonomous cars, allowing them to recognize a stop sign or differentiate a pedestrian from a lamppost, and is the basis for voice control in smartphones, tablets, and TVs.
What is deep learning?
Deep learning is a subfield of machine learning, an area of AI that uses algorithms to teach a machine to perform a certain task. While traditional machine learning algorithms are linear, deep learning algorithms are stacked in a hierarchy of increasing complexity and abstraction.
To understand deep learning, imagine a baby whose first word is dog. The child learns what a dog is, and what it is not, by pointing to objects and saying the word dog. The parent says: “Yes, that is a dog” or “No, that is not a dog”. This principle can be applied to the most basic machine learning.
As the child keeps pointing at objects, it becomes more aware of the characteristics that all dogs have. What the child unknowingly does is elaborate a complex abstraction, the concept of a dog, by building a hierarchy in which each level of abstraction is created with the knowledge gained from the previous layer of the hierarchy.
Similarly, through deep learning, a machine is also able to construct abstract concepts. For this it needs a large amount of data and substantial processing power.
Deep learning vs. machine learning
Machine learning algorithms leverage structured and labeled data to make predictions, which means that specific features are defined from the input data to the model and organized into tables.
This doesn’t necessarily mean that it doesn’t use unstructured data; it just means that if it does, it usually goes through some pre-processing to organize it into a structured format.
Deep learning eliminates some of the data pre-processing that is normally involved in machine learning. These algorithms can ingest and process unstructured data, such as text and images, and automate feature extraction, removing some of the reliance on human experts.
For example, say we had a set of photos of different pets and wanted to categorize by “cat”, “dog”, “hamster”, etc. Deep learning algorithms can determine which features (e.g. ears) are most important in distinguishing one animal from another. In the most basic machine learning, this hierarchy of features is established manually by a human expert.
Then, through the processes of gradient descent and backpropagation, the deep learning algorithm adjusts, allowing it to make predictions about a new picture of an animal with greater accuracy.
Applications of the technology
Deep learning technology is part of our lives, but in most cases it is so well integrated into products and services that users do not realize the complex data processing that takes place in the background.
Algorithms can be used to identify dangerous patterns that indicate possible fraudulent or criminal activity in banks, in chatbots used in customer service, in virtual assistants such as Apple’s Siri, Amazon Alexa, or Google Assistant, and even with applications in medicine, helping medical imaging specialists and radiologists produce diagnoses in less time.
How to work with Deep Learning?
Bachelor’s degrees in Computer Science can provide an overview of the concept of AI and deep learning. However, to better understand the concept and seek a position in the field, it is important that the professional take a graduate course. Researching the subject is another exciting possibility.
Source: UPM, IBM, SAS.
