Artificial Intelligence and Machine Learning

Abstract— Artificial Intelligence commonly used to account

A. AI type-1: Based on Capabilities
for the use of various levels of details in a given representation
language or the ability to change from one level to another 1. Weak AI or Narrow AI:
while preserving useful properties. Machine Learning an
application of artificial intelligence where available Narrow AI is a type of AI which is able to perform a
information is used through algorithms to process or assist the dedicated task with intelligence. The most common and
processing of statistical data. currently available AI is Narrow AI in the world of
Artificial Intelligence. Narrow AI cannot perform
Keywords— Natural language processing (NLP), database, beyond its field or limitations, as it is only trained for
computer vision, neutral network, overfitting. one specific task. Hence it is also termed as weak AI.
Narrow AI can fail in unpredictable ways if it goes
I. INTRODUCTION beyond its limits. Apple Siris a good example of
Artificial Intelligence is the capability of a computer Narrow AI, but it operates with a limited pre-defined
system to mimic human cognitive functions such as learning range of functions. IBM's Watson supercomputer also
and problem-solving. Through AI, a computer system uses comes under Narrow AI, as it uses an Expert system
maths and logic to simulate the reasoning that people use to approach combined with Machine learning and natural
language processing. Some Examples of Narrow AI are
learn from new information and make decisions formatting.
playing chess, purchasing suggestions on e-commerce
Machine learning is programming computers to optimize site, self-driving cars, speech recognition, and image
a performance criterion using example data or past recognition.
experience. We have a model defined up to some
parameters, and learning is the execution of a computer 2. General AI:
program to optimize the parameters of the model using the
training data or past experience. The model may be General AI is a type of intelligence which could
predictive to make predictions in the future, or descriptive perform any intellectual task with efficiency like a
to gain knowledge from data.[1] human. The idea behind the general AI to make such a
system which could be smarter and think like a human
by its own. Currently, there is no such system exist
II. DEEP LEARNING which could come under general AI and can perform
any task as perfect as a human. The worldwide
A. Deep Learning in ML researchers are now focused on developing machines
Machine Learning means computers learning from data with General AI. As systems with general AI are still
using algorithms to perform a task without being explicitly under research, and it will take lots of efforts and time
programmed. Deep Learning uses a complex structure of to develop such systems.
algorithms model on the human brain. This enables the 3. Super AI:
processing of unstructured data such as documents, images,
and text. Super AI is a level of Intelligence of Systems at which
machines could surpass human intelligence, and can
B. Deep Learning in AI perform any task better than human with cognitive
Deep learning drives many artificial properties. It is an outcome of general AI. Some key
intelligence applications and services that improve characteristics of strong AI include capability include
automation, performing analytical and physical tasks without the ability to think, to reason, solve the puzzle, make
human intervention.[2] judgments, plan, learn, and communicate by its own.
Super AI is still a hypothetical concept of Artificial
Intelligence. Development of such systems in real is
still world changing task.
B. Artificial Intelligence type-2: Based of functionality
1.Reactive Machine
Purely reactive machines are the most basic types
of Artificial Intelligence. Such AI systems do not
store memories or past experiences for future
III. TYPES OF ARTIFICIAL INTELLIGENCE actions. These machines only focus on current
Artificial Intelligence can be divided in various types, there scenarios and react on it as per possible best action.
are mainly two types of main categorization which are IBM's Deep Blue system is an example of reactive
based on capabilities and based on functionally of AI. machines. Google's AlphaGo is also an example of
Following is flow diagram which explain the types of AI. reactive machine.
2. Limited Memory
Limited memory machines can store past experiences or understand the images, such as the shape & size of the tail
some data for a short period of time. These machines can of cat and dog, Shape of eyes, colour, height (dogs are taller,
use stored data for a limited time period only. Self-driving cats are smaller), etc. After completion of training, we input
cars are one of the best examples of Limited Memory the picture of a cat and ask the machine to identify the
systems. These cars can store recent speed of nearby cars, object and predict the output. Now, the machine is well
the distance of other cars, speed limit, and other information trained, so it will check all the features of the object, such as
to navigate the road. height, shape, colour, eyes, ears, tail, etc., and find that it's a
cat. So, it will put it in the Cat category. This is the process
3. Theory of Mind of how the machine identifies the objects in Supervised
Theory of Mind AI should understand the human emotions, Learning.
people, beliefs, and be able to interact socially like humans.
B. Unsupervised Machine Learning
This type of AI machines are still not developed, but
researchers are making lots of efforts and improvement for Unsupervised machine learning is different from the
developing such AI machines. Supervised learning technique; as its name suggests, there is
no need for supervision. It means, in unsupervised machine
4. Self-Awareness
learning, the machine is trained using the unlabelled dataset,
Self-awareness AI is the future of Artificial Intelligence. and the machine predicts the output without any
These machines will be super intelligent, and will have their supervision.
own consciousness, sentiments, and self-awareness. These
machines will be smarter than human mind. Self-Awareness In unsupervised learning, the models are trained with the
AI does not exist in reality still and it is a hypothetical data that is neither classified nor labelled, and the model
concept.[3] acts on that data without any supervision. The main aim of
the unsupervised learning algorithm is to group or categories
the unsorted dataset according to the similarities, patterns,
and differences. Machines are instructed to find the hidden
patterns from the input dataset.

Let's take an example to understand it more preciously;

suppose there is a basket of fruit images, and we input it into
the machine learning model. The images are totally
unknown to the model, and the task of the machine is to find
the patterns and categories of the objects.

So, now the machine will discover its patterns and

differences, such as colour difference, shape difference, and
IV. TYPES OF MACHINE LEARNING predict the output when it is tested with the test dataset.

Machine learning is a subset of AI, which enables the C. Semi-Supervised Learning

machine to automatically learn from data, improve
performance from past experiences, and make predictions. Semi-Supervised learning is a type of Machine Learning
Machine learning contains a set of algorithms that work on a algorithm that lies between Supervised and Unsupervised
huge amount of data. Data is fed to these algorithms to train machine learning. It represents the intermediate ground
them, and on the basis of training, they build the model & between Supervised (With Labelled training data) and
Unsupervised learning (with no labelled training data)
perform a specific task. algorithms and uses the combination of labelled and
unlabelled datasets during the training period.
A. Supervised Machine Learning
Although Semi-supervised learning is the middle ground
Supervised Machine Learning is based on supervision. It
between supervised and unsupervised learning and operates
means in the supervised learning technique, we train the
on the data that consists of a few labels, it mostly consists of
machines using the "labelled" dataset, and based on the
unlabelled data. As labels are costly, but for corporate
training, the machine predicts the output. Here, the labelled
data specifies that some of the inputs are already mapped to purposes, they may have few labels. It is completely
the output. More preciously, we can say; first, we train the different from supervised and unsupervised learning as they
are based on the presence & absence of labels. To overcome
machine with the input and corresponding output, and then
the drawbacks of supervised learning and unsupervised
we ask the machine to predict the output using the test
learning algorithms, the concept of Semi-supervised
dataset.
learning is introduced. The main aim of semi-supervised
Let's understand supervised learning with an example. machine learning is to effectively use all the available data,
Suppose we have an input dataset of cats and dog images. rather than only labelled data like in supervised learning.
So, first, we will provide the training to the machine to Initially, similar data is clustered along with an
unsupervised learning algorithm, and further, it helps to the environment, moves of an agent at each step define
label the unlabelled data into labelled data. It is because states, and the goal of the agent is to get a high score. Agent
labelled data is a comparatively more expensive acquisition receives feedback in terms of punishment and rewards.
than unlabelled data.
Due to its way of working, reinforcement learning is
We can imagine these algorithms with an example. employed in different fields such as Game theory, Operation
Supervised learning is where a student is under the Research, Information theory, multi-agent systems.
supervision of an instructor at home and college. Further, if
that student is self-analysing the same concept without any A reinforcement learning problem can be formalized
help from the instructor, it comes under unsupervised using Markov Decision Process (MDP). In MDP, the agent
learning. Under semi-supervised learning, the student has to constantly interacts with the environment and performs
revise himself after analysing the same concept under the actions; at each action, the environment responds and
guidance of an instructor at college. generates a new state.

D. Reinforcement Learning
Reinforcement learning works on a feedback-based process,
in which an AI agent (A software component) automatically
explore its surrounding by hitting & trail, taking action,
learning from experiences, and improving its
performance. Agent gets rewarded for each good action and
get punished for each bad action; hence the goal of
reinforcement learning agent is to maximize the rewards.

In reinforcement learning, there is no labelled data like REFERENCES

supervised learning, and agents learn from their experiences
only. [1] www.splunk.com/en_us/data-insider/ai-and-machine-learning.html
(references)
The reinforcement learning process is similar to a human [2] www.ibm.com.
being; for example, a child learns various things by [3] javatpoint.com
experiences in his day-to-day life. An example of www.javatpoint.com/types-of-machine-learning
reinforcement learning is to play a game, where the Game is