Tuesday, April 23, 2019
Sunday, March 10, 2019
Artificial Intelligence- Machine Learning and Deep Learning
AI is nothing but the capability of machine to imitate intelligent human behavior. AI is achieved by mimicking a human brain by understanding how it thinks, how it learns and works while solving any problem. AI includes various areas of specification:
-Game playing
-Expert System e,g number recognition from the number plate
of speedy car
-Natural Language Processing
-Neural Networks
-Robotics etc
Artificial Intelligence is huge term, it includes Machine Learning and Deep Learning. Basically it is the super-set of ML and DL.
Machine learning is a type of AI which provides computers with the ability to learn without being explicitly programmed. Here we do not need to define all the steps of conditions like any other programming application. however we have to train the machine on a data set(training data), viz large enough to create a model which helps to take decisions based on its learning capability.
For Example: We've to train a machine to determine the species of flower; for that we give it flower data set which contains various characteristics along with respective species like; flower name, sepal and petal length, species name, color,number of sepals and petals, life span etc. using this data set machine will create a model which can be used to identify the flower. Next time we pass some characteristics of flower, machine will display the name of the flower by analyzing these characters.
Although it sounds good and easy to implement but it has a limitation. ML is not capable of handling high dimension data where it has to deal with bulk of input and output data, processing such data becomes complex and resources get exhausted. This is termed as Curse Of Dimensionality. So the crucial problems with huge data cannot be solved with ML.
Another big challenge for traditional ML is Feature Extraction. For Example object recognition, handwriting recognition. These tasks cannot be accomplished by ML, clearly we needed a better approach and here came the concept of Deep Learning.
Deep learning is capable of handling high dimensional data. It is also efficient in focusing on the right features on its own that is feature extraction. Deep learning is implemented through Neural Network which is an Artificial Neuron also called as Perceptron and motivation behind the neural network is Biological Neuron. Deep learning attempts to re-engineer a human brain and studies basic units of brain called Brain cells/Neuron.
DL is very vast and cannot be narrated in one paragraph so I am going to write on Deep learning with more exhaustive details in my coming blogs
-Game playing
-Expert System e,g number recognition from the number plate
of speedy car
-Natural Language Processing
-Neural Networks
-Robotics etc
Artificial Intelligence is huge term, it includes Machine Learning and Deep Learning. Basically it is the super-set of ML and DL.
Machine learning is a type of AI which provides computers with the ability to learn without being explicitly programmed. Here we do not need to define all the steps of conditions like any other programming application. however we have to train the machine on a data set(training data), viz large enough to create a model which helps to take decisions based on its learning capability.
For Example: We've to train a machine to determine the species of flower; for that we give it flower data set which contains various characteristics along with respective species like; flower name, sepal and petal length, species name, color,number of sepals and petals, life span etc. using this data set machine will create a model which can be used to identify the flower. Next time we pass some characteristics of flower, machine will display the name of the flower by analyzing these characters.
Although it sounds good and easy to implement but it has a limitation. ML is not capable of handling high dimension data where it has to deal with bulk of input and output data, processing such data becomes complex and resources get exhausted. This is termed as Curse Of Dimensionality. So the crucial problems with huge data cannot be solved with ML.
Another big challenge for traditional ML is Feature Extraction. For Example object recognition, handwriting recognition. These tasks cannot be accomplished by ML, clearly we needed a better approach and here came the concept of Deep Learning.
Deep learning is capable of handling high dimensional data. It is also efficient in focusing on the right features on its own that is feature extraction. Deep learning is implemented through Neural Network which is an Artificial Neuron also called as Perceptron and motivation behind the neural network is Biological Neuron. Deep learning attempts to re-engineer a human brain and studies basic units of brain called Brain cells/Neuron.
DL is very vast and cannot be narrated in one paragraph so I am going to write on Deep learning with more exhaustive details in my coming blogs
Friday, March 8, 2019
Robotics, Automation Vs Artificial Intelligence
Previously we peruse introduction to Artificial Intelligence, crux was that AI is the capability of machine to imitate the intelligent human behavior and it is achieved by mimicking the human brain by understanding how it learns and works while solving the problems.
Since there are various emerging crafts of technology that people may sometimes get disoriented about various terminology like AI, Robotics, Automation and use these terms interchangeably like a triune.
Here I will make it clear, differences and similarities about these three terms in order to get rid of this mystification.
Automation is described simply by two words- Manual Input and Rule-Based output.
It means you set a formula or an Algorithm which applies some mathematical substitutions and transformations on the given data and gives us output. Every time we give an input, it applies same rules and conditions and begets the output. For example if there is a company, customers can mail them any time( that is the manual input), if the employee is free only then he interacts otherwise an automatic reply is send "we will assist you later" or something like that, and that is the rule based output which depends on the condition whether the employee is free or not. This is how it works, no change in rules no improvement in functionality or efficiency.
In case of Artificial Intelligence, the more data we give to our machine, the more experienced and intelligent it gets and every time it produces an output with more efficiency then the previous output. AI improvises exactly the way humans does. Automation applies rules to data and produces output, but AI not only applies rules but learns from the data, memorizes it and apply the rules more accurately next time.
Robotics includes AI, robotics involves many different things with lot of mechanics like sensors, activators, maths and their is a part of AI in it which makes it intelligent. Robotics also includes the moment(which actually makes it entirely different from other fields). Moment is not always the displacement of machine, in many industries there are robot arms that works and carries things from one place to another. This motion of machine at correct positions/locations involves mathematics and AI and lot of other things. Robotics is not same as AI but in robotics part of AI is applied in it.
So we can't use the three terms interchangeably all three are different from each other from their logic and specification point of view.
Since there are various emerging crafts of technology that people may sometimes get disoriented about various terminology like AI, Robotics, Automation and use these terms interchangeably like a triune.
Here I will make it clear, differences and similarities about these three terms in order to get rid of this mystification.
Automation is described simply by two words- Manual Input and Rule-Based output.
It means you set a formula or an Algorithm which applies some mathematical substitutions and transformations on the given data and gives us output. Every time we give an input, it applies same rules and conditions and begets the output. For example if there is a company, customers can mail them any time( that is the manual input), if the employee is free only then he interacts otherwise an automatic reply is send "we will assist you later" or something like that, and that is the rule based output which depends on the condition whether the employee is free or not. This is how it works, no change in rules no improvement in functionality or efficiency.
In case of Artificial Intelligence, the more data we give to our machine, the more experienced and intelligent it gets and every time it produces an output with more efficiency then the previous output. AI improvises exactly the way humans does. Automation applies rules to data and produces output, but AI not only applies rules but learns from the data, memorizes it and apply the rules more accurately next time.
Robotics includes AI, robotics involves many different things with lot of mechanics like sensors, activators, maths and their is a part of AI in it which makes it intelligent. Robotics also includes the moment(which actually makes it entirely different from other fields). Moment is not always the displacement of machine, in many industries there are robot arms that works and carries things from one place to another. This motion of machine at correct positions/locations involves mathematics and AI and lot of other things. Robotics is not same as AI but in robotics part of AI is applied in it.
So we can't use the three terms interchangeably all three are different from each other from their logic and specification point of view.
Wednesday, January 23, 2019
My Reviews On Artificial Intelligence
Artificial Intelligence is one of the most outstanding craft of computer science field and it is a sheer fluke for me to try my hands in it.
To describe AI three beautiful and organic words came in my mind- Think, Analyse and Make Decisions. These three are what comprises of intelligence.
All of us learn at every single instance of life, every time we dabble in anything (good or bad) we learn from it. so, learning from past experience and applying it in future decisions is Intelligence.
For example babies have their own instincts (that they subconsciously shift with time), they always make attempts to play with electrical or electronic items and their parents or elders always push them back, but at least once they make their successful attempt and they happen to get electric shock. Yes, that is the most objective phase of learning for them. From that on wards they automatically stop to go near those items because they learn it practically that electrical items have something harmful in it. This is an intelligence, the more you experience and the more you do, the more you improvise- A very organic thing in humans.
Now when we try to import this intelligence in machine we call it "Artificial Intelligence".
For a machine past experience is basically "data", more we give it more it learns and gets better and better. For every thing to do, again we have three words to describe: Capability, Task and Method.
AI is the capability of machine to imitate(task) the intelligent human behavior an it is achieved by mimicking the human brain by understanding(method) how it learns and work while solving problems.
Artificial Intelligence includes various areas of specification like Expert Systems, NLP, Neural Network, Game playing and Robotics etc.
John McCarthy(Father of AI) states: "AI is the science and engineering of making intelligent machines, especially intelligent programs".
Artificial Intelligence is accomplished by studying how human thinks, learns, decides and work while trying to solve problems, so that when the same or even more complicated problems will be given to our machines they will follow the same organic procedure of humans. This studying does include various things that we have to know in exhaustive details so that the crux can be applied to our beloved machines i,e
- knowledge of computer science(including some programming language)
-Mathematics
-Biology( to understand the structure of human brain)
-Neural Science(working of human brain)
-Psychology(psychological affects on human intelligence)
-Sociology(sociological affects on human intelligence)
-Philosophy : curiosity- can machine think like human?
To describe AI three beautiful and organic words came in my mind- Think, Analyse and Make Decisions. These three are what comprises of intelligence.
All of us learn at every single instance of life, every time we dabble in anything (good or bad) we learn from it. so, learning from past experience and applying it in future decisions is Intelligence.
For example babies have their own instincts (that they subconsciously shift with time), they always make attempts to play with electrical or electronic items and their parents or elders always push them back, but at least once they make their successful attempt and they happen to get electric shock. Yes, that is the most objective phase of learning for them. From that on wards they automatically stop to go near those items because they learn it practically that electrical items have something harmful in it. This is an intelligence, the more you experience and the more you do, the more you improvise- A very organic thing in humans.
Now when we try to import this intelligence in machine we call it "Artificial Intelligence".
For a machine past experience is basically "data", more we give it more it learns and gets better and better. For every thing to do, again we have three words to describe: Capability, Task and Method.
AI is the capability of machine to imitate(task) the intelligent human behavior an it is achieved by mimicking the human brain by understanding(method) how it learns and work while solving problems.
Artificial Intelligence includes various areas of specification like Expert Systems, NLP, Neural Network, Game playing and Robotics etc.
John McCarthy(Father of AI) states: "AI is the science and engineering of making intelligent machines, especially intelligent programs".
Artificial Intelligence is accomplished by studying how human thinks, learns, decides and work while trying to solve problems, so that when the same or even more complicated problems will be given to our machines they will follow the same organic procedure of humans. This studying does include various things that we have to know in exhaustive details so that the crux can be applied to our beloved machines i,e
- knowledge of computer science(including some programming language)
-Mathematics
-Biology( to understand the structure of human brain)
-Neural Science(working of human brain)
-Psychology(psychological affects on human intelligence)
-Sociology(sociological affects on human intelligence)
-Philosophy : curiosity- can machine think like human?
Subscribe to:
Posts (Atom)