Scribd
Upload
62 views14 pages

Assignment # 1

This document contains an assignment for a machine learning course. It discusses labelled and unlabeled data. Labelled data has tags or labels assigned to raw data samples, which is used for supervised learning tasks like image recognition. Unlabeled data does not have labels assigned, and is used for unsupervised learning where the model must identify clusters or patterns. The document also contains multiple questions about machine learning concepts like K-nearest neighbors classification and adjusting K values locally versus globally.

Uploaded by

Fasih Dawood
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
62 views14 pages

Assignment # 1

This document contains an assignment for a machine learning course. It discusses labelled and unlabeled data. Labelled data has tags or labels assigned to raw data samples, which is used for supervised learning tasks like image recognition. Unlabeled data does not have labels assigned, and is used for unsupervised learning where the model must identify clusters or patterns. The document also contains multiple questions about machine learning concepts like K-nearest neighbors classification and adjusting K values locally versus globally.

Uploaded by

Fasih Dawood
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 14

Machine Learning For Data Science

Assignment # 1

Muhammad Fasih Dawood


Roll # 20k-1395

Question # 1
Labelled Data is type of data where the meaningful significant labels, classes or tags are already assigned
to that raw data. In simple words, it’s also a collection of samples with one precise tag or meaning, for
example that you have dataset of patients, and each patients having different symptoms, and the data
consisting of specified result of Cancer, so upon the results, the tag can easily be set for each patient and
identify, whether the patient having positive cancer test or negative cancer test. Another, so that data of
labelled is a group of various samples, those group of samples already tagged with characteristics,
classifications and properties of samples. It’s specially used while developing the machine learning
models on the aspects of supervised learning, it’s also used to recognize the image, that also called image
recognition, it decides according to tagged labels of image, what actually the image belongs for. It’s also
used to ascertaining the raw data including text files, videos, images, etc.

Unlabeled data is a type of data, in which the pieces of raw data is not properly tagged with labels,
recognizing properties, features or classifications, it’s also used in numerous forms of Machine Leaning.
That unlabeled data is belongs to the unsupervised models of machine learning, and it doesn’t have
labelled data. Let’s take an example of unlabeled data and how it’s classify, through the example of
patients, so here the clusters must be defined for each patient i.e. patients having different symptoms and
also has similarities based on data. In case, the model identifies the patient symptom through clusters
those are made up according to properties, resemblances of patients.
Machine Learning for Data Science Assignment # 1

Question # 2
Machine Learning for Data Science Assignment # 1
Machine Learning for Data Science Assignment # 1
Machine Learning for Data Science Assignment # 1
Machine Learning for Data Science Assignment # 1
Machine Learning for Data Science Assignment # 1
Machine Learning for Data Science Assignment # 1
Machine Learning for Data Science Assignment # 1
Machine Learning for Data Science Assignment # 1
Machine Learning for Data Science Assignment # 1
Machine Learning for Data Science Assignment # 1
Machine Learning for Data Science Assignment # 1
Machine Learning for Data Science Assignment # 1

Question # 4
In this paper, they took out the method of KNN Classifier and kept perspectives; KNN’s classification
one of the method which is used to be on several stimulating features, it’s even good for generalization
and easily implemented. Although, it’s typically keeps ability towards even out-perform and matching as
well, more classy and multifaceted techniques. But the problem is identified about to approaching to fix
the values that has to be suitable, and so the suitable value can be classified through Cross-Validation
(CV). It’s improbable that value which is same might be ideal aimed at the entire interplanetary traversed
by the set of training data. So the apparent result can be seen as that dissimilar various constituencies of
the space which is featured, and that definitely be requiring the diverse k-values cause of prototyping
distributions, and that could be different. Whereas, this status quo of instances of query is quietly unalike
from condition of two-classes residing near between boundaries of two classes. They’ve approached the
robust method to setting up the k-value locally and then to very prototype the k-value potentially to be
different and get the greatest k-value through heightening a standard comprising locally and the effects of
dissimilar k-values in the near region of the prototype which is considered globally effects to the
prototype. So it concise that they training stage became faster and robust through this method and
approach they’ve proposed and the stage of testing having the same complexity as usual method of KNN.
According to the method results, experimental design shown that modest approach of KNN which is
standard can expressively out-perform for both problems of class imbalanced and standard; reflected on a
set which is larger of various kind of problems. The author’s already publicized experimental efforts that
their technique which does better rather than those three variants of KNN and also standard one. they
have summed up with work of future would aim at this method to be parallelizing and for learning of
optimal local values of K, using algorithms those are evolutionary evolving.

You might also like