top of page
  • Writer's pictureAdmin

Deep learning is coming to us.

Ramon Murguia Deep Learning

Ramon Murguia Difference between techniques

Deep learning is a term that we might also known as deep structured learning or hierarchical learning, is part of machine learning methods based on learning from data representations, as opposed to task-specific algorithms. There is a difference among machine learning, Artificial intelligence and big data analytics, statistics and some other new trends on technology that are helping, shaping and changing our capacity to process information today and will impact our near future.

Today we will talk about deep learning as a part of this world and how is contribution is helping in some areas of science and some industries to transform the way of working and their capacity to process and learn faster than ever.

In deep Learning there are 3 ways to process and manage data. The information can be supervised, semi-supervised or unsupervised. To better understand the implications let us figure it out how each work. In supervised learning, the program is stocked with inputs that are values or labels and output and machines are challenged with finding a function or equation that approximates this behavior in a generalizable fashion (i.e. finding a customer that might not pay you, based on previous events). The output could be a class label (in classification) or a real number (in regression)-- these are the "supervision" in supervised learning. Semi-supervised learning case, involves function estimation on labeled and unlabeled data. This approach is motivated by the fact that labeled data is often costly to generate, whereas unlabeled data is generally not. The challenge here mostly involves the technical question of how to treat data mixed in this form. In the case of unsupervised learning, is when you have no labeled data available for your exercise and the algorithm does everything and gives you the outcomes. So deep learning, working with other algorithms, can help you classify, cluster and predict. It does so by learning to read the signals, or structure, in data automatically.

Now imagine that, with deep learning, you can classify, cluster or predict anything you have data on either supervised, semi supervised or unsupervised, for example: excel spreadsheets data, social media behavior, web information or even more complex things such as images, video, sound, text and DNA, time series (touch, stock markets, economic tables, the weather, earthquakes). If the previous list gives you a hint, that is, anything that humans can sense and that our technology can digitize. With deep learning, we have basically the ability to behave much more intelligently, by accurately interpreting what's happening in the world around us with software because we have multiplied our ability to analyze what's happening in the world by many times.

Taking this extremely complex but heavily simplified background description, on how data is managed. Let me start with some applications that are being used today:

Sound has many uses, here you have already your voice as a security password in some banks, also user experience analysis, user interaction and that is chatbots that are answering machines that can answer and have conversations with you on any subjects you program machines to perform. Moreover, sentiment analysis, research in a short time the web to understand what is being said about your brand, competition or anything you might want to know and in some industries such as automotive and aircraft by deep learning with sound you can evaluate flaws on performance to a level that humans might not.

Ramon Murguia Deep Learning approach

Image has a broad implication, the most well know today is facial recognition, image files are 1s and 0s that computers can analyze and predict and cluster together to recognize everything within a picture. with this simple explanation imagine you want to search for pictures on the web or social media for a particular aspect (this for security, crime detection, fraud, etc). Automotive technology will also benefit form this, since you can upload to learn a massive amount of images for road recognition, aviation as well. Now form the telecom industry or governments might be able to cluster persons based on images they take on their phones. Some other benefits are the ability to restore colors in B&W photos and videos, Pixel image restoration to generate a high quality image or describing photos.

Video deep learning is being used today for motion detection. Here you can see it in security cameras, home smart things, airport security, human behavior, motion on streets, warehouse to mention some as you can see in the video below.

Now some other usage of deep learning for you to get a broader visualization of the potential and usage are:

Translation, saving whales and classifying Plankton, reading text in images and videos, self-driving cars, robotics, voice generation, music composition, restoring sound in videos, transferring artistic style from famous paintings, automatically writing Wikipedia articles, math papers, computer code and even Shakespeare, handwriting, predicting demographics and election results, predicting earthquakes, healthcare to mention some of the usage being tested out today.

Ramon Murguia Deep Learning Examples

40 views0 comments

Recent Posts

See All
bottom of page