gracylayla (4) [Avatar] Offline
In 1950’s Artificial Intelligence emerged making the Information Technology more feasible to humans and human related tasks happen more easily and accurately. In 1980’s, a new subset emerged out of Artificial Intelligence called as Machine Learning making use of human inputs, it analyses and gives output more accurately. Now comes another subset of AI via Machine language called as “DEEP LEARNING”

The technology is fundamentally altering the way we live, work, and communicate akin to the industrial revolution – more specifically making us to think less and providing information within no time.

The shortest Explanation:
* Artificial Intelligence: Computer system(s) that replicates human intelligence.
* Machine Learning: Makes computers to perform a self-learning.
* Deep Learning: Algorithms attempting to model high level abstractions into data to determine a high level meaning.

Let’s Go Deeper into Machine Learning vs Deep learning:

Machine Learning:

It is the subset of Artificial intelligence.
It covers all types of data science algorithms and pattern recognitions.
These algorithms can work on low performance computers without GPU as there is no need to store any data.
They breakdown the problem into pieces and solve them individually and then combine it.
They required features to be identified and then coded manually
They have a defined set of rules that can be understandable.
They take more time in testing phase rather than training the model.
It uses various types of automated algorithms and learn to model functions and predict future actions from data.
These algorithms are directed by Data analysts
It allows a systems to recognize patterns on their own and make future predictions.
It requires someone to continuously code or analyze data to solve a problem and predict a result.
They can quickly be applied on facial, speech, object recognitions, translation, and many other.

Deep learning:

It is a subset of Machine learning.
It covers neutral networks covering network layers and parameters.
These algorithms required high performance computers with GPUs and TPUs with huge storage power.
They solve the problem end-to-end.
They try to identify high-level features from low-level features.
They have mathematically complicated rules which are difficult to understand.
The volumes of data would be in millions as it deals with big data.
The output can be anything- numerical, free-form elements like test, image videos or sound.
It uses neural networks that passes data through many processing layers to interpret data about the features and their relationships.
It uses some machine learning techniques connecting neural networks that simulate human decision-making.
They can be applied on image , video, sound and text recognitions etc.

It can be expensive and requires huge datasets to train itself.
It is a bit automatic.

What is TensorFlow?

TensorFlow is a multipurpose open source so2ware library for numerical computation using data flow graphs. It has been designed with deep learning in mind but it is applicable to a much wider range of problems. .

But what does it actually do? TensorFlow provides primitives for defining functions on tensors and automatically computing their derivatives.