Original Source Here
Get the speed and higher potential growth with deep learning
-Why does deep learning so much disruptive potential?
-What deep learning do to our business?
-How does we combine deep learning with reinforcement learning, computer vision, Neural language processing?
So we can say “Deep learning is the family of machine learning based on Artificial neural network. Deep learning is the discipline within Artificial Intelligence that teaches computers how to make prediction base on raw data”.
As all we know the most suffisticated and difficult to study human brain. Human brain is very complex in itself and having million, billion numbers of neurons. So Deep learning is mimic the human brain inside the machine. In deep learning there is no predefined frameworks. Suppose supervised or unsupervised learning what we do … we try the algorithm and framework one by one and put the best fit. Human brain consist of neural networks now the question arrives what is neural network.? …it is artificial construct designed human brain to mimic inside the machine We can say the machine think and behave like human brain working.
We can take the example of a newborn baby how the baby starts learning. How to walk how to talk the baby learn slowly with time.
There is a drawback of deep learning it required much more data like thousands of images too much data set for learning and take decisions.
Deep learning was started in 80s. The methods used in 80s has disappeared now and we’ll try again to build deep learning concepts, neural networks And now we are leading them with 100 and thousands of examples and we can get the results insights within an hour or within a minute it depends on your system capability.
See see many of you think that if deep learning able to learn by self or through by data so why do we need engineers and deep learning expert?
Let me answer you the deep learning consists of input layer hidden layer and the output layer this is the basic structure of deep learning. So in this so in this structure we cannot predict the number of hidden layer. We can define the number of hidden layer till then our data not trained so we cannot predict the number of hidden layer. The hidden layer consists of neurons and there are different way we can construct neuron the method is called architect and the architect is very complex process. The neuron consist of weight and memory. To overcome from the complex architect and the technical process we can say we need experts, programmers and deep learning experts.
when you perform they transform part of neural network you have lot of supervised data. Let’s take a example of we have lots of images around 10,000 more than 10,000 images of dog and cats. put these images in input layer put it the vector inside and get the output if we’re getting the output from the cat image the output getting get cat.It means we have trained our model accurately. But suppose what happened when the input is dog and our model tell us this is a cat it means we have to correct our model and go deep inside the data. This technique is called back propagation.
After many many iterations you will get the neural network tailored to predict whether an image is a cat or dog.
What happen if we combine deep learning with the other area of artificial intelligence.
Other area of artificial intelligence like computer vision, deep learning, reinforcement learning, neural language Processing. The above video of artificial intelligence we are required for training our model. The deep learning is not required always after using this method if we’re going to use the deep learning then we increase the power of artificial intelligence.
If we combine deep learning with computer vision then it is called deep computer vision.
If we combine deep learning with reinforcement learning then it is called deep reinforcement learning.
If we combine deep learning with neural language processing then it is called deep neural language processing.
After this combination the drawback of the most is the complex computation.
What deep learning do to our business
Now we will discuss most real life used cases of deep learning.
1. DL in CANCER DETECTION by Goggle.
2. DL in RECOMMENDATION ENGINEER by Spotify.
3. DL in MINERAL EXPLORATION Buy gold spot discoveries.
4. DL in VISUAL EFFECT by digital domain
5. DL in ANTI MONEY LAUNDRING by Ayasdi.
6. DL in CYBERSECURITY by deep instinct.
7. DL in PRODUCTIVITY TRACKING by doxel.
8. DL in facial recognition by Amazon rekognition.
9. DL in REAL STATE PRICES by Zestimate.
10. DL in LOAN APPROVAL by Zest finance.
Trending AI/ML Article Identified & Digested via Granola by Ramsey Elbasheer; a Machine-Driven RSS Bot