Deep Learning and How exactly does it work?

Deep Learning and How exactly does it work?

Source

Deep learning is basically a subset (application) of machine learning . Machine learning works with simple concepts as well as pattern recognition , deep learning is based on ann (artificial neural networks) which was actually designed on the basis of an actual human 🧠 brain. With the increase in level of bug data analytics the complexity of ann has also been incredibly considerably. Deep learning can be applied in speech recognition, language translation etc.

Source

✔️How does deep learning work?
Neural networks can be compared to the neurons present in the human brain.Nodes are connected to each other since they are in individual adjacent layers. Just like a single neuron in the brain it receives hundreds of signals from other neurons , in ann the signal travels through various nodes and assigns corresponding weights so heavier weighted nodes will have more effect on adjacent nodes. The final layer compares all the inputs and gives an output. To perform this very powerful 🖥️ computer hardware are needed. For many years neural networks was not the meta since we didn't have powerful graphic processing units . If there are more data sets the more accurate results we get.

✔️Where is deep learning seen in action?

Source
There are many deep learning libraries in the current era such as Tesorflow , DSSNTE by amazon etc.
Image recognition in Facebook where the computer auto tags our friends by detecting thier faces in a image , here face detection is applied with the help of deep learning!
Alexa ,Siri ,google assistant use deep learning for speech/language recognition.
Windows teams , youtube translate the audio into real time cc (caption) without any lag. Even mails nowdays automatically go into spam . All this is only possible easily with the application of deep learning.
Google's deepmind " Alpha Go " which on just being trained for 6 hours beat an former world champion in the game of go.

Self driving cars nowdays are being trained on deep learning based models where they can detect and avoid muntiple objects at once. Neural networks are getting more accurate by the day in predicting weather , stock prices etc.

Source

Deep learning application as well as models will definitely keep increasing over time as higher graphic processing units are created. The predictions will be much more accurate and dependable .

Hope you learnt something new! Let me know your thoughts in the comments ^^