Deep Learning Technology.
                     Deep Learning Technology
Introduction
• Deep learning is a technique of machine learning, teaching computers to do what comes 
naturally to humans: learn by example. 
• Deep learning is a key technology behind driverless cars, allowing them to recognize a stop sign, 
or distinguish a pedestrian from a lamppost. This is the secret to voice control in consumer 
devices such as phones, laptops, televisions, and hands-free orators.
Source: Pixabay
How does deep learning attain such 
Impressive results?
1. Deep learning involves large quantities of labelling data. For example, development of 
driverless automobiles requires millions of images and thousands of hours of video. Deep 
learning calls for significant computational power. 
2. High-performance GPUs have a parallel architecture which is effective for profound learning.
Source: Pixabay
Examples
1. Automated Driving 
2. Aerospace and Defense 
3. Medical Research 
4. Industrial Automation 
5. Electronics
Source: Pixabay
How Deep Learning Works??
• Many deep learning approaches use architectures of the neural network which is why deep 
learning models are also referred to as deep neural networks. 
• The word "deep" typically refers to the amount of layers concealed within the neural network. 
Standard neural networks comprise only 2-3 hidden layers while deep networks can have up to 
150.
Source: Mathworks
CNN or ConvNet
• One of the most common forms of deep-neural networks is known as neural convolution (CNN 
or ConvNet). A CNN combines learned features with input data, and uses 2D convolution layers, 
making this architecture suitable for 2D data processing, such as images. 
• CNNs eliminate the need for manual extraction of the feature, so you don't have to identify the 
features used to classify images. The CNN operates by directly extracting features from images.
Source: Wikipedia
Difference between Deep Learning and 
Machine Learning?
• Specialized form of machine learning is deep learning. A machine learning workflow begins 
with the manual extraction of relevant features from images. 
• They then use the features to create a model that categorizes the objects in the image. With a 
deep learning workflow, the necessary features are extracted from the images automatically. 
Deep learning also carries out "end-to - end research."
Source: Towards Data Science
Training Methods
1. Training from Scratch
2. Transfer Learning
3. Feature Extraction
4. Accelerating Deep Learning Models with GPUs
Source: Pixabay
Deep Learning Applications
• Pre-trained deep neural network models can be used by conducting transfer learning or 
extraction of features to rapidly apply profound learning to your problems. 
• Some models available for MATLAB users include AlexNet, VGG-16, and VGG-19, as well as 
imported coffee models (for example, from Caffe Model Zoo) using importCaffeNetwork.
Source: Pixabay
Deep Learning with MATLAB
• MATLAB supports the deep learning. MATLAB also offers advanced toolboxes for working with 
machine learning, neural networks , computer vision and autonomous driving, with software 
and functions for handling massive data sets. 
• MATLAB helps you to do deep learning without being an expert, with only a few lines of code. 
Get started quickly, build and visualize models, and deploy models to servers and devices that 
are embedded.
Source: Twitter
THANK YOU!!!
Source: Pixabay