1K Blog Marathon: Day 58
Yesterday was the first and only time that I searched for “Mountaineering Tent” in Google. Today, I can see other variations of tent, rough road boots, mountaineering backpacks and even survival tools! Then after few searches on programming, the ads were now showing web hosting offers! I know they are relevant, but how impressive these ads at targeting clients!
It’s not magic – aside from the fact that it is pure algorithmic learning. Deep Learning, to be exact. Do you know what really Deep Learning is? Are you aware that every day you are exposed to this algorithm, and that you, aside from being a human, is now a data input for a machine to learn?
Deep Learning, meaning?
Okay, to understand Deep Learning, let’s get to the top of the pyramid.
We have Artificial Intelligence (AI), the intelligence, reasoning and response of a machine. It has many sub-branches, one is Machine Learning.
It is the application of AI that helps computer system to learn by giving input data. The computer system’s task is to devour that data, learn from it, and improve its internal system and judgment on its own. It has a sub-branch, which is Deep Learning.
This is the practice of making a machine more intelligent by imitating the human brain’s function, called “Artificial Neural Networks”. This is the part of my blog where readers like you will either choose another one of my blog, or just open Facebook ^_^
To throw away the technicalities, Deep Learning is when a software gets as many data as possible, learn from it, update its own algorithm, and providing sound and more accurate output over the time. By imitating how our brain works, its pattern recognition and judgment today becomes superior to yesterday.
Self-driving cars, face and pattern recognition, even search suggestions – Deep Learning is everywhere.
Payment thru Smiling
In China, shoppers are paying their goods by means of “Face Recognition”. They don’t have to bring money or credit card. They don’t even need to use their smartphones for mobile payments.
World Chess Master Defeated by Computer
Garry Kasparov, a World Chess Champion battle with Deep Blue, IBM’s Artificial Intelligent.
JARVIS, Iron Man’s AI assisting him, is maybe not far from real. With the Virtual Assistants like Siri, Alexa and Cortana, Deep Learning are helping them improve their services, at least virtually.
Self-driving vehicles, such as Trucks, cars and drones, are also utilizing Deep Learning in their system, so they can know what a stop sign is, perform maneuvers when on a sharp curve, or even avoid collision.
I bet you already encountered one. You can talk to one in Facebook Messenger of a Group or Company, which will send you either relevant answer, long answer with Call-To-Actions link, or even an “I don’t know what you are talking about” response. You know its bot when it sounds like a bot.
What’s mind blowing?
Although Deep Learning is based on the learning pattern of a human brain, being a system on itself, it has some advantages and disadvantages of its own.
Robots, unlike humans, can learn faster and share this learning across a network of robots easily. In humans, 2 heads are better than one, is also applicable in Deep Learning. Once interconnected, the rate of learning is multiplied by the robots involved. For example, 10 Deep Learning Bots will function at once, sharing the learning of each and using it at once. We can say that they all learn 10 times faster. Compared to 10 humans, they can only learn as fast as their brain and memory can. They may share information, but their differences in absorbing and interpreting information may be a problem.
Humans, when given even limited examples, can learn it and even come up with a more creative solution. This is not in the case of bots. Because they follow a strict algorithm of Input-Learn-Output, they must require a very large amount of sample data, and their learning factor may increase slowly or part by part. They can’t come up with logical answer based on limited information, or else they will provide misjudgment. Also, large amount of data needs a large amount of storage – from Gigabytes to even Terabytes of data storage. They also needed more computing power, which can be expensive.
Technology has reached far and wide, and taken us to Mars. Deep Learning has many significant improvements that aid humans, and to learn more about ourselves. Let’s just take a cautious action on every step for humanity, else it will be our last foot print.
“AI is likely to be either the best of worst thing to happen to humanity.”Stephen Hawking
All images are owned by their respective owner. Source for every images are supplied.