Now let’s have a look at one of the buzz words in the recent time i.e. A.I. This word brings different emotions to a different set of people many are afraid of it and many are totally into it to unlock the mysteries of nature. here the thing how can A.I. just unlock some of the mysteries of nature, many think that A.I when leading to a greater extent can help us to find the unbreakable mystery of nature “THE INTELLIGENCE”. In my sense A.I is something where a software is created using our intelligence to make the host look that it contains intelligence one example of A.I in our daily life is google assistance it is so perfect that it can create the remainder for you when you receive a mail like you have something on someday, it can also see your daily travel style and make the source place as home and destination as work and also it gives updates on traffic condition in the routes.
The main features that an A.I has to possess is it should be capable of handling a large amount of different type of data, as the data produced has been increased at a greater rate , it should also be capable of learning and improve itself by the results, it must also be changed to the conditions around like it has to continuously check the environment and when required has to react ,it should also be capable of predicting things beforehand.
Machine learning is one of the derivatives of A.I, Machine learning is where we provide machines to learn all by itself without being explicitly programmed that is, A computer program is said to learn from some experience E with respect to some task T and some performance measure P, if its performance on T, as measured by P, improves with experience E.
Learning can be classified into two ways
- Unsupervised learning
- Supervised learning
Supervised learning is where we predict the output by mapping it with the input using the help of predefined input-output pairs. The main task of this type of learning is finding the perfect style of the pairing pattern to produce the results to the future inputs the two main methods of this process is regression and classification.
Regression is a process of finding the pattern between the variables.
Let’s consider the above graph as the input and output graph regression is a best possible shape that fits the dataset and which helps to find the outputs for future inputs. The best possible shape can be
From this, we can project the input value to the line and then check the output value by projecting it into the output axis.
To find the shape we need to pull the hypothesis hθ (x), where
hθ (x)= θ0+θ1x+θ2x2+ ……+θnxn
this equation gives the required shape for the pattern.
Here the values of θ are unknown to find it we use a cost function
J(θ)=1/(2m) Σi = 1 to m(hθ (x)i-y^i)2
This is the cost function which is used to find the value of theta now the thing is that we have to find the optimal value of theta for that we have many algorithms one of them is gradient descent, in which we have to change the value theta until we reach the optimal solution.
θ i := θ i – α * ∂(j(θ))/∂θ
Alpha is learning rate.
This repeats until we find the optimal solution.Then how we know we reached an optimal solution? This can be found by knowing the change when the change decreases and increases at a point the point where it increases is optimal point. Alpha place a vital role in reaching an optimal point if alpha is too large it crosses the optimal point and makes it tough to get to the optimal point if it is too small it takes too much time to reach the optimal point.
In this way, we find the hypothesis which gives the equation of shape.
Classification is the process of finding the category for the given input. for example, if we give a text the process of finding whether the text is good or bad can be done by classification. Given a hypothesis similar to regression we can filter it by making values between 0 to hypothesis line to a class and hypothesis line to into another class.
The dashed line in the above figure is the hypothesis line and the right side is one class and the left side is the another.
but here a problem arises what if h(θ) is more than 1 or less than less 0 and also the greater outliers can also cause a greater effect on the hypothesis this may cause errors. This can be eliminated by using the sigmoid function, sigmoid function pulls the values within the range.
Let θ0+θ1x+θ2x2+ ……+θnxn =y
to find the θ we need cost function. This cost function is similar to the one of regression except that the value of hθ (x) is different. To get the optimal value we use gradient descent which is quite similar to one of regression.
Unsupervised learning is the type of learning where the machine makes the pattern by using datasets without any label responses. The main method used in this learning is cluster analysis.
Cluster analysis is grouping the similar type of entities. This makes us know the underlying pattern of the entities. This can be achieved by using k-means clustering and hierarchical clustering.
In this way the machine undergoes learning. As many think A., I can be an apocalyptic thing my assumption is uncontrolled A.I can be a destructive one, let’s remember
“If machine makes a mistake it is not the mistake of the machine it is the mistake of the human who coded it.”
Let’s assume that you are in a depressed state of mind because of a series of events and your mental health is quite low, you don’t have any human available to contact at that particular instance, what would you do to feel good??
Well, we have chatbots which are intelligent enough to enact human-like behaviour in the most natural way by having an emotionally uplifting conversation with the user by reacting and reciprocating to the texts of the user. These types of bots are generally deployed over social media or on Mental health websites where humans who are emotionally not feeling well would or who would have suicidal tendencies text their health condition or text about how they feel to the bot, and the bot analyses the texts and based on the previous records it would reciprocate accordingly by making the human feel comfortable and less stressed.
This world is undergoing enormous amounts of changes every day, which is resulting in the complexity and to solve this complexity we have Artificial Intelligence. As a quote states;
“It’s better to do one task in a perfect manner rather than doing hundreds of tasks with imperfection”
AI is not only making lives easier but also helping in evolving humans by reducing the hard work and increasing the time constraint for humans to do the smart work. Artificial Intelligence, when embedded into devices, can do miracles by predicting and solving logical and analytical problems to an extent. However, human interference is required at a point of time to correct the decisions made by Artificial Intelligence. Even after so many advances in the field of Artificial Intelligence the question still remains the same,
“Is it a New Beginning or an Inevitable End?”
Chatbots and Artificial Intelligence
A research article on “Artificial Intelligence” made possible with the guidance and help of,
Head of Department(Information Technology)
SreeNidhi Institute of Science and Technology,
Researched and Published by,
Sai Srujan, IT-F4, 15311A12I4, SNIST
*Not for reproduction
Reblogged this on The Research Society and commented:
Research article by Sai Srujan, 15311A12I4, IT-F4 (2015-2019) on “The Art of Thinking/Artificial Intelligence”.