Mastering DL - Ep0: A sneak-peak to Deep Learning🧐️

Why Am I making this introduction?😄️

Sometimes it is important to have written backup or notes of what you have learned and your thoughts. I do not tend to talk a lot or be in public debates, so this is my way of contributing with little knowledge to everyone.

Deep Learning is a promising technology that can radically transform the world we live in and it deserves all the attention it is getting from researchers, AI-first businesses, and media alike.

Please do not say that deep learning is just adding a layer to a Neural network, and that's it, magic! Nope. I'm hoping that after reading this, you have a new perspective of what DL is.

Prerequisites before entering DL⏪️

You should be familiar with Applied Math & Machine Learning Basics + Interest;

• Linear Algebra
• Probability Theory
• Statistics
• Numerical Computation
• ML Basics

In case you start DL without first completing these prerequisites:

What is Deep Learning?🤔️

Deep learning is a subset of Machine learning that uses mathematical functions to map the input to the output. These functions can extract non-redundant information or patterns from the data, which enables them to form a relationship between the input & output. This is known as learning, and the process of learning is called training.

Modern deep learning models use Artificial Neural Networks or simply neural networks to extract information. These neural networks are made up of simple mathematical functions which can be stacked on top of each other and arranged in the form of layers, giving the sense of depth, that's why the term deep learning.

Neural Networks🧠

The Neural network is the heart(Dil ki dhadkan🤭️) of deep learning models, and it was initially designed to mimic the working of the neurons in the human brain.

Which looks like this;

Neural networks enable us to learn the structure of the data or information and help us understand it by performing tasks such as clustering, classification, regression or sample generation.

DL vs ML, Why Deep learning?🧐️

Let me tell you that Deep learning is more powerful than traditional Machine learning algorithms. Generally, machine learning is like shallow learning it is effective on smaller datasets. On the other hand, Deep learning becomes extremely powerful as we increase the size of the dataset. To demonstrate the level of interest in Deep Learning, here is the Google trend for the keyword: DL can draw accurate inferences and learn complex patterns from data on its own. It is so OP that it can even process unstructured data, such as text corpora, social media activity, and so on. Furthermore, it can also generate new data samples and find anomalies that machine learning algorithms and human eyes can miss. Deep Learning shines when it comes to complex problems such as image classification, natural language processing, and speech recognition.

How Deep Learning Works?😲️

Neural networks have multiple layers of interconnected artificial neurons which are stacked together. Each of these neurons has a simple mathematical function - commonly a linear function - that conducts information/data extraction and mapping.
There are three layers to a deep neural network: the input layer, hidden layers, and the output layer. The data is fed into the input layer first. Each neurone in the input layer injects the data and transfers it to the next layer, i.e., hidden layers. These hidden layers extract features from the input layer and transform them using a linear function.

These layers are called hidden layers because each neuron has unknown parameters such as weights and biases; these layers add random parameters to transform the data, each of these yields a different output. This output is then passed to the final layer, the output layer, which classifies, predicts or creates data depending on the task.

This overall process is called Forward Propagation.

Then comes the Backward Propagation or Backpropagation. It's an algorithm like Gradient Descent which calculates errors by taking the difference between the predicted output and original/expected output.

This error is then fine-tuned by travelling backwards through the layers, fine-tuning hyper-parameters like weights and biases of the function.

Therefore, the process of forwarding and backward propagation both allow a neural network to reduce error and achieve high accuracy in a specific task. With each epoch/iteration, it becomes gradually more accurate.
Bingo!🤩️ Now you know what Neural networks are!

Types of Neural networks 📶️

Artificial Neural Networks(ANN)

Convolutional Neural Networks(CNN)

Recurrent Neural Networks(RNN)

Generative Adversarial Networks(GAN)

PS: We will be covering all types of Neural networks in detail under this DL Series.

Applications of Deep Learning🏎️

Finally, here are some of the real-life applications of deep learning.

• Healthcare⚕️
  • Medical image analysis
  • Surgical robotics
• Autonomous Industry🤖️
  • Self-driving cars (Tesla♈️)
  • Smart cities
• Agriculture👨‍🌾️
  • Plant disease and pest detection
  • Livestock monitoring
  • Robot picking
• Gaming🎮️

Future of Deep Learning♉️

"The art of forecasting is predicting what will happen and then explaining why it won't!"😆️

I'm not sure how to forecast the future. But what I can say is what I see happening and what could happen. The acquisition of DeepMind Technologies by Google bodes well for worldwide marketers, so look out for DL in the future!

Any questions, reach me out✌️

Thanks for reading this weird introduction to Deep Learning. I hope it helped get you started in this fascinating field, or even just discovering something new.

• Email: saurabhpowar1823@gmail.com
• LinkedIn: https://www.linkedin.com/in/saurabh-powar-72b4701b4/

Phir Milenge...! 😅️🤘️