Scribd
Upload
13 views3 pages

Ai Answers

Uploaded by

partharya269
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
13 views3 pages

Ai Answers

Uploaded by

partharya269
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

SUBJECT SPECIFIC SKILLS

18. Two approaches in the modelling stage of AI project cycle include :


Rule based approach : In this approach , certain rules for the identification of datasets is
defined by the developer . The rules are encoded in the machine , and the machine
works on the basis of these rules , without deviating from them . The machine does not
function properly for an entirely new data set as the learning is static .
Learning based approach : In this approach , the rules are not defined by the developer ,
and the machine learns from the testing data upon which it is tested . This approach
helps the machine to learn dynamically , and to tackle new kinds of datasets with
varying patterns and trends .
20. Supervised learning refers to a type of learning based approach in which the dataset
is labelled by the developer , and therefore the machine is able to classify it easily
without observation of any trends . It requires external supervision . Its two types include
:
i) Classification : This type of data problem helps us to classify two datasets and
categorize / label them into the category in which they belong . For instance : Students
which are pass / fail can be easily done by this data problem . This data problem works
on discrete data sets .
ii) Regression : This type of data problem is accurate for making predictions upon the
basis of the previous observations or previous data . In this data problem , the machine
is able to predict the next set of values on the basis of the past information . It is used
for stock market prediction or salary prediction .
19. Unsupervised learning refers to a type of learning based approach in which the
dataset is not labelled by the developer , and the machine is left on its own to identify
the patterns or the trends associated with it . It does not require any external supervision
. Its two types include :
i) Clustering : This type of data problem helps us to identify common objects from a set
of different objects and label them into certain categories by identifying new patterns or
trends which fall into one category .
ii) Association : This type of data problem helps us to deal with large number of
databases by identifying certain rules and associating our observations with it . For
instance : In a market , a large group of people who buy X also tend to buy Y , this type
of scenario can easily be dealt with association data problem .
21. Reinforced learning refers to another type of learning based approach in which the
agent recognizes and learns from the environment around it . It is not fed up with rules ,
or left on its own to identify patterns , rather it sees the environment around it . This
type of data problem does not require any kind of external supervision , and it is based
on the reward system .
22. A neural network or an artificial neural network (ANN) is a stimulation of the human
brain . It involves the connection of various layers in an ANN to receive the input ,
process it and transfer the output . Its various types include :
1. Perceptron – It is a single layered neural network which does not have any hidden
layer . This means that input is directly sent out as output .
2. Feed forward network – It is a collection of various perceptrons , though 1 hidden layer
is present in it , however it is not connected to the neural network through nodes .
3. Radial basis network – This neural network is a collection of an input layer , hidden
layer and output layer where a node connects both the input layers . It is used for
functional approximations and predictions of numerical values .
4. Deep feed forward network – This neural network is a complex collection of an input
layer , many hidden layers and an output layer . The signal is received at the input
layer , which is processed in the hidden layer , and sent out as output layer .
5. Recurrent neural network – This neural network consists of an input layer , a recurrent
cell layer and an output layer . In this neural network , various hidden layer cells
combine to form a recurrent cell , so it takes time for the signal / input to reach every
recurrent cell and the output to emerge out of it .
23. Advantages and disadvantages of a neural network
Advantages of a neural network Disadvantages of a neural network
 It has high fault tolerance  It needs to emulate human
 It can process large amounts of data intelligence
 It can be implemented in any  It requires high processing time
application  A neural network requires time for
 It is highly accurate if provided with its training
the right data

24.
Structured data Unstructured data
 It is a collection of well defined and  It is a collection of random and
labelled data sets with same data unlabelled dats sets with
features distinguishing data features
 It is easier to work on structured  It is quite difficult to work on
data unstructured data

25. Data science is a branch of artificial intelligence which deals with data , not
particularly data , but data implemented and integrated with statistics and mathematical
concepts like probability , finance , counting etc. Four applications of data science
include :
 Aggregator websites : Aggregator websites , also known as price comparison
websites , extensively use data science , as they need to collect data from all the
websites on the internet , and compare their prices and present it to the user . This
involves dealing with a lot of data in terms of retail price and quantity of the
furnished good .
 Genetical Science : Genetics refers to the study of inheritance of characteristics in
organisms . Data science helps us to record and store this information in
computers . It also helps us to analyse our genome , which in turn helps us to
identify pre – natal characteristics and genetic disorders . It also aids in genetic
engineering of plants and animals .
 Online recommendation : Data science is extensively used for the recommednation
purporses in apps such as netflix and prime video . Whatever content we watch , it
asks for rating , and as per our rating , the recommendation is given to us . Thus ,
the data is stored and the recommendations are given to us
 Airplane routing : Airplane and civil aviation extensively uses data science for its
daily purposes . Data science is used for checking the aircrafts in air , air traffic ,
trajectories of different aircrafts and mapping the position of aircraft with
longitudinal coordinates . It all uses data science .
26. Four challenges associated with data science are :
 Data science team in offices are not given adequate attention , and is often
neglected
 It requires highly precise and accurate data for generating results
 Means of storing large amount of data is vert expensive , and companies cannot
pay for it
 Data science often is not used to its potential , and the research is not properly
implemented
27. Various tasks associated with CV (Computer Vision) are :
Single object Multiple Object
Classification : It involves the labelling of Object detection : Object detection
the specific object in the captured image , involves detecting various objects in the
which is categorised by the system image . For example : A system will
Classification + localisation : It involves identify the specific images of dog , cat
the proper labelling and outlining the and a fish from an image containing all
desired object which was classified by the the three
system . It is often seen as the step Instance segmentation : It involves the
preceeding the object detection differentiation or segmentation of various
detected objects into their own
categories , as the machine draws a rough
outline around them.

28. Four applications of CV include :


 Self – driving cars
 Social media and entertainment
 Biometrics and face recognition
 Online search engine : Google lens
29.

You might also like