Deep Learning Demystified
Uploaded by
Perets ArnaudDeep Learning Demystified
Uploaded by
Perets ArnaudTatap Perets
Sep 22, 2022 · 3 min read · Listen
Deep Learning Demystified
From Kevin Ku at [Link]
Nowadays, AI is revolutionizing the way we live. From the use of Siri or Alexia on
our smart phones to facial recognition patterns, AI has proven to be truely useful in
simplifying our daily lives. However, AI is broad, and as we get narrower, the
strength of the AI algorithm gets better.
Convert web pages and HTML files to PDF in your applications with the Pdfcrowd HTML to PDF API Printed with [Link]
Relationship between AI, ML and DL
Artificial Intelligence (AI) refers to the simulation of human intelligence in
machines that are programmed to think like humans and mimic their actions.
Inside AI we have a subset known as Machine Learning (ML), which is the use and
development of computer systems that are able to learn and adapt without following
explicit instructions, by using algorithms and statistical models to analyse and draw
inferences from patterns in data. Then inside ML exists a subset called Deep
Learning (DL) which is based on artificial neural networks with representation
learning. So all deep learning models are machine learning models, but the reverse
is not always true.
Deep learning algorithms are better than machine learning algorithms because they
perform feature extraction and classification simultaneously, saving processing
time and gaining in efficiency.
How does Deep Learning work?
Convert web pages and HTML files to PDF in your applications with the Pdfcrowd HTML to PDF API Printed with [Link]
Fundamental structure of a deep learning model
Deep Learning is based on neural networks, which imitates the neurons of the
human brain. There is an input layer, some hidden layers, and an output layer.
1 The input layer
The input of the model can be an image, a sound, a video, an object, or any other
element which the model is to learn. The input node takes in information in a
numerical form. The information is presented as an activation value where each
node is given a number. The higher the number, the greater the activation. Based on
the strength of the connections, weights are given to each node in the input layer.
The weights are the inputs multiplied with a random number.
2 Hidden Layers
The weighted sum from the input layers are the entries of the first hidden layer.
This form a bias. When all the bias have been formed they go through an activation
value that determines which bias is qualified to move to the next layer, and so on.
Examples of an activation function includes the sigmoid, the tangent, the
hyperbolic, the rectifying linear activations.
Convert web pages and HTML files to PDF in your applications with the Pdfcrowd HTML to PDF API Printed with [Link]
3 The Output Layer
The number of nodes at the output layers represent the final classification groups;
usually there are two nodes at the output layer. The last hidden nodes will undergo a
final activation layer screening, and will be fed into the output layer. At the output
layer, a probability is designed to know where the result will get.
Conclusion
Several modules exists with which one can create and train his deep learning
model. They include Tensorflow, Keras, PyTorch, Caffe,and others. Knowing how to
build, train, and launch a deep learning model is a must-know-how skill every
serious tech enthusiast should consider possessing.
Artificial Intelligence Machine Learning Deep Learning
More from Tatap Perets Follow
Telecom/ Network Engineer, AI Enthusiast. Let’s connect
[Link]
Love podcasts or audiobooks? Learn on the go with our new app.
Try Knowable
Recommended from Medium
Aditya Pawar
Trends In Artificial Intelligence
Sachin Vyas in Thoughts & Collections
Singularity, Artificial ‘general’ Intelligence & ‘No free lunch’ in Machine Learning
Convert web pages and HTML files to PDF in your applications with the Pdfcrowd HTML to PDF API Printed with [Link]
