Basically deep learning is very important part of machine learning which is developed to bring machine learning closer to most important area i.e. Artificial Intelligence. From past years almost before 2015, AI was just a part of our imagination. It was just a concept since 2015 on which researchers were working. now deep learning has come into existence.
After 2015, AI started exploding. This becomes possible because of increase in technology. Various technologies are being introduced since 2015 which makes AI possible somehow. Especially after the introduction of GPU, machine learning lead to a boom. After GPU was introduced, we become able ton do multiple tasks at same time with very high speed and low processing time. It made our work more faster and easier at very cheap rate. It provides us huge storage through which we can store huge data very easily at low cost. Basically deep learning is a network of interconnected neural network units.
AI vs machine learning
Basically deep learning is a part of machine learning and it was introduced to bring us closer to AI. First question arises in everyone’s mind is what is AI? What is the difference between machine learning and AI? Let me tell you about AI and how machine learning differs from AI i.e. machine learning vs Artificial Intelligence.
Both machine learning and AI are often used interchangeably, specially for handling big data. But these are not same. Lets discuss some points about machine learning vs artificial intelligence.
Artificial Intelligence is a part of machine learning. Artificial Intelligence is based on making a machine intelligent which carried out its tasks based on algorithms in an intelligent manner. It is not complete machine learning it’s a part of that. On the other hand machine learning is building an ability of a machine to get trained from a set of data and learn from them, and make future predictions or we can say generate results according to that data processing.
We train the systems using neural networks and try to make them capable to think like humans. These neural networks play a vital role in training a machine to think like humans. The Neural networks works with a series of algos through which our system can recognise all possible patterns as our brain works. Through neural networks we can extract the meaning of complicated data.
On the other hand deep learning is also a subset of machine learning which is known as another deeper level of machine learning. It is a deep neural network which includes various layers .These layers are used to process the data and to divide it in various classes and will work like decision trees. It will train huge data in a simple manner and will make data meaningful for us and this data is used for future predictions.
Some important applications of deep learning:
Lets discuss about various applications of deep learning which tells us how it will change our life. Different types of technology we developed using it, how much powerful it is. Some of it’s applications are as follows:
1.Changing Colors of photos:
Using it now we are able to colour our black and white photos. Nowadays we all use filters on our photos using various effects. That technology becomes possible just because of it. Basically deep learning actually finds the pattern that naturally occurs in photos for example grass is natural green and using these patterns it will give colours to our pictures.
2. Speech to Text Conversion:
It’s the one of the best technology introduced using deep learning. Nowadays we don’t need to write a message we just simply speak and machine will write it for us. This becomes possible only because of deep learning. Deep learning will recognize the pattern between the voice and words through which it makes technology like speech to text conversion possible.
This is also one of the most successful technology developed using deep learning. Now we don’t need to hire persons for customer care services. We can use chat robots which will do reply to customers for us. And Moreover, a person can handle a single client at one time but a ChatBot can handle number of clients at same time. Using this, we train a dataset which covers all replies and conversation between various people. It will recognise pattern of replies according to a question and will do future predictions or we can say give replies accordingly.
We all had heard about it. One of the vast technology among all. This all becomes possible just because of it. An automatic car which runs itself without any human intervention. This car will detect objects automatically which includes people and road signs too. This technology will actually tell us the potential of deep learning. How it will change our life.
This technology is opposite of speech to text conversion. This is also a great technology introduced using it. We can also generate voice from the text which becomes possible only through deep neural networks.
6. Music Composition:
Another application of deep learning is music composition. Well known Example of Francesco Marchesani who trained the computer to compose music like my favorite classical composer Chopin. After the computer learns the patterns and statistics that are unique to the music of Chopin, it creates a completely new piece.
7. Adding Sounds to Silent Movies:
One of the coolest application of deep neural networks is automatically adding sound to silent movies. Through deep learning models which is built with pre-recorded sounds of database to select a sound for future predictions of what happened in scene. It is one of the best technology introduced by deep learning.
8. Automatic Machine Translation:
Another well known technology introduced by deep learning using neural networks is automatic machine translation system. Nowadays we have systems which will translate our language into various other languages. This helps us in communicating with all kinds of people in world. This amazing technology will become possible just because of deep learning.
I hope this post will be helpful for all of you. Now we all know how deep learning can change our life. How much potential it have. We can use AI in such a way which can work for humanity and development. But we should not forgot that every thing has its negative impacts too. We all know how powerful deep learning is. We can make many impossible concepts possible through deep neural networks. But it is important to remember that we should use technology like deep learning in a positive manner which will be beneficial for all. I hope this post will give you idea about present and future of deep learning.
12 thoughts on “How deep learning changes your life in unexpected way”
Deep learning is subfield of machine learning whose algorithms and concepts are designed in a similar fashion the human brain is working. One answer to the growth of this part occurs because today high computational computers are available and also huge amount of data is available which makes training easy and accurate. Deep learning is magical and fantastic at supervised learning. Tensor flow is the result which indicates how fast deep learning is growing. Huge investment by Google in its Google brain project and same for amazon and Microsoft. Amazon and Microsoft is spending huge amount of money in this field. This all indicates how important deep learning is. Deep learning is a huge network that requires huge amount of data and high processing speed. It finds many application.
Thanks john for your point of view with respect to tensor flow. You have good knowledge about project Google brain. Deep learning has been an integral part of all the multinational and fast growing companies. This reflects the importance and great demand of it. Huge amount of money is spend in doing the research and development in this domain.
If I look at google api or clarifai api, all are using deep learning in some respect. Today there are a lot of field which are using deep learning to develop novelity.
Deep learning is a machine learning framework. Finds application in speech, language, vision, playing games like Go. On the other hand it requires huge amount of data to get trained which could be done by GPU (Graphically Processing units). Further feature learning is an intrinsic property of deep learning and hence a lot of time is saved in manually feature extraction or selection step. In conclusion, Deep Learning has a great advantages vs. shallow learning, because deep nets can learn very complex functions which we even hardly understand.
Thanks Sam for your valuable information that you have shared. I am quite sure that though deep learning process is quite expensive, it produces good accuracy.
Machine learning is an important area which needs to be understood. But shifting the paradigm towards deep learning is the matter of concern here as you have written beautifully about it. I am trying to boil down some information that I have in this comment. From my knowledge machine learning is used when the dataset is quite small and deep learning is used when the training and testing data set is quite high and simple CPU’s cannot handle it. Secondly in deep learning, learning is done using weights. One layer adds the weight to the input provided to them and others layers manipulate these weights. Based on these weights, output is generated.
Thanks mandeep singh for your comment. This reflects you can differentiate well between both machine learning and deep learning. Yes true concepts you have explained.
First of all congratulation for writing such a blog. The question that I am asking is out of the context but still I have in my mind so I am asking. I have made a small app using python programming language but I got an assignment to convert it into the .exe file.
I came across your article here and hope that you may help me out.
First of all Glbaat welcomes you for reaching us here. It is great to write to you. We appreciate your question and donot worry will provide you the best possible answer as we can
Please follow the below both link to get your answer: One is you tube video and another is the well documented git hub repository
How are you. I saw that you write on deep learning. I had visited your other websites such as
I would like to ask you that I am running the code using the keras at the frontend and tensorflow at the backend. I need to change tensorflow to theano. Could you suggest me how to change it. I am using ubuntu.
Great to hear from you such question. Hope i will provide you the answer that can help you out.
First of all please go to the home folder and type ls -a. Type this command in the terminal and will list all the files in the home directory. Please look for the file .keras and open it using your favourite editor. I use nano to open it. You may use sublime, gedit or vim.
Once the file is opened, please change the backend from tensorflow to theano.
You are done.
Thanks for the answer. There is another question which comes to my mind.
I need to draw the diagram/graph for the decision tree which is the machine learning classification as well as regression model. But when I run the code to draw the tree, it reflect me the error of pydot and graphviz?
Please suggest me how to resolve this issue???/
pydot and graphviz are needed to plot the image for the decision tree. Please visit https://askubuntu.com/questions/917030/how-to-install-pydot-and-graphviz to look for the solution. If you want me to provide you the solution please run the below code in the terminal. You can open terminal using Ctrl – Alt + T
I hope that your problem will be resolved.