联系方式

  • QQ:99515681
  • 邮箱:99515681@qq.com
  • 工作时间:8:00-23:00
  • 微信:codinghelp2

您当前位置:首页 >> Java编程Java编程

日期:2020-12-11 11:21

Distributed Computing and Problem

Solving: 7 5hp

Assignment 1

3th November 2020

Inneh?ll

Part 1 Rules ............................................................................................................................................. 3

Timelimit............................................................................................................................................. 3

Aid....................................................................................................................................................... 3

Submitting ........................................................................................................................................... 3

Presentation ......................................................................................................................................... 3

Grades.................................................................................................................................................. 4

For the grade 3p:.............................................................................................................................. 4

For grade 4p: ................................................................................................................................... 4

For grade 5p: ................................................................................................................................... 4

For the grade U:............................................................................................................................... 4

How do I refer to a source? ................................................................................................................. 5

Part 2 Hands on ....................................................................................................................................... 6

Aim...................................................................................................................................................... 6

The Assignment, the coding!............................................................................................................... 6

Data ................................................................................................................................................. 6

For grade 3 .......................................................................................................................................... 6

For grade 4p: ....................................................................................................................................... 8

For grade 5p: ....................................................................................................................................... 9

References ............................................................................................................................................. 9

Part 1 Rules

Timelimit

Se information in learn.

Aid

This assignment can be solved individually or in pairs but I recommend in pairs. Watch out for

plagiarism.

Plagiarism is the shortness to hand in on someone else's work as their own. Obviously, if it is a

couple's work, then you will have the same solution, but both names need to be on the solution. In

many cases, plagiarism, this can happen unconsciously, and that is why you have to read the

instructions.

Submitting

When submitting apply all options:

Choose to whether write in English or Swedish (don’t mix).

Name your zip file with your username and the grade your aiming for. For example: h16kriha_4.zip in

learn. The submission time will be automatically registered when you submit. Late submissions can’t

grade a higher grade then G(3p) so be careful with the deadlines!

When in a group all need to do individually submission because there is a oral presentation that is

individual even when working in group.

The submission will include one zipfile containing: A report (with answers, sources and a use case

UML), a coding project and a video/videolink with the oral presentation.

Presentation

Everyone in the group needs to:

Add their name to all the documents (in code add all the names as comments, there should not be any

questions regarding who wrote the code). THIS IS VERY IMPOERTANT. Hand in the submission

into the folder in learn before deadline has passed. Late submissions will not be graded higher then G

(3p) and will be examined when the teacher has time for it, it is not guaranteed that it will be graded in

the same time as the ones that handed in “in time”.

The code need to be runnable, make sure to test and check that all you needed to have included is

included in the zip file.

Also everyone in the group need to individually do a presentation that should not succeed 15

minutes (MAX 15 minutes!). You can choose a recording software but its not allowed to have

background music in the recording, your voice needs to be clear. Hand in the link to your video in

your submission. The oral presentation should include:

? Explanation of the code and keywords for each grade, marked with a * in the part 2 Hands on

? Explanation of the solutions and if you hade any problems (how did you overcome them).

? Do a real time run of the code and explain the results that is showing.

Grades

This hand in have four grades U (failed) 3p, 4p and 5p. All the tasks need to be complete and finished

to get a point/grade, there is no subpoints, its Boolean.

For the grade 3p:

All tasks for grade 3p need to be solved, completely solved. Its Boolean, either it works or it does not.

? Create a report where you (can be done in a group):

o Describe the data that you choose to use.

o Discuss the problems and solutions that has come up. (was it hard or easy, did you

find other ways to solve it?)

o Explain the program, what does it do and use at least one source to your explanation,

example a peer-reviewed article or a book.

o Your answers and your source need to be relevant to the task. (see more detailed

information of what is needed in the report and code in part 2 Hands on)

o All your answers and comments must make sense, the teacher should not need to

reread to understand what you wrote.

o Create a use-case UML diagram, se example in figure 2, and include it in the report.

? Solve the coding task, se more information in part 2 Hands on (can be done in a group).

? Do a individual oral presentation (a video).

For grade 4p:

All the tasks in grade 3p need to be completed AND all for grade 4. Its Boolean, either it works, or it

does not. Also, the hand in is submitted in time in learn, se deadlines and time limits in learn.

? Everything for the grade 3 and…

? Do the extra tasks in the coding (se Part 2 Hands on for more information)

? In the report write about the “new” agent from the coding task for grade 4p in Part 2 Hands

on. And also explain Recall and F1 score.

? You need to refer to at least two different sources and at least one needs to be peer-reviewed.

For grade 5p:

All the tasks in grade 3p and 4p need to be completed AND all for grade 5. Its Boolean, either it

works, or it does not. Also, the hand in is submitted in time in learn, see deadlines and time limits in

learn.

? Everything for the grade 3p and 4p and…

? In the report add a section where you write about the algorithm: Naive Bayes for classification.

? Do the extra tasks in the coding (se Part 2 Hands on for more information)

? You need to refer to at least three different sources and at least two needs to be peerreviewed.


For the grade U:

If any of the demands for grade 3p is not fulfilled.

How do I refer to a source?

Literature references that supports your text can be taken from:

Chapter one and two in the course literature Russel and Norvis Atrificial Intelligence: A modern

Approach (3,ed), 2010

Peer-review articles from ACM Digital Library, log on with your school username and password.

Peer-review articles from ScienceDirect (Elsivier Journals), log on with your school username and

password.

Use APA-reference style. It can look like this:

… the Otsu thresholding method (Otsu, 1979) was used ….

And in the reference list it can look like this:

Otsu, N. (1979). A threshold selection method from gray-level histograms.

Automatica, 11(285.296), 23-27

OBS! All articles you quoted must be fully downloaded in this assignment, saved and enclosed in

your submission. (This to simplify and fast the correction procedure)

Part 2 Hands on

Aim

You have come far into your education and has experiences in how to study and how to do

problemsolving. This task can be a challenge but its also a good training for how far you have come.

Some parts come from courses earlier in the program, some parts from this course and some part could

be new. The assignment is fairly open so hence your creativity will be in focus, there is not only one

way to solve this.

The aim is:

Gain insight in how agents communicate with each other (including transferring data) based on a

asynchronous massage passing paradigm. Hence, you should implement the communication part by

making use of the FIPA Agent Communication Language (ACL) Specifications. An ACL Message

contains a number of fields including these four:

? The sender

? The receiver

? The communication Act

? The message content

Grimshaw, 2010.

You should after this know how agents can send data between each other and the act of transmitting

data.

The Assignment, the coding!

Data

Use one or both of these datasets for your machine learning.

? winequality-white.csv - https://archive.ics.uci.edu/ml/machine-learning-databases/winequality/winequality-white.csv

? Students knowledge about direct current systems -

https://archive.ics.uci.edu/ml/datasets/User+Knowledge+Modeling

Often, you must clean and setup the data according to the predetermined rules, which will

suit the computer system. E.g since Excel files carry a lot of meta data, you might want to

transform all files to *.csv, you might want to determine which decimal separator you want

to use, i.e ',', or '.' etc etc.

For grade 4p and 5p in the target column in each file, you must add the string "target". This string

should later be recognized by an agent, who determines if this dataset should be conducted by using

supervised or unsupervised learning.

For grade 3

Create a multi-agent system MAS (Using JADE) where in at least two agents are active.

See description of the suggested architecture in figure 1, where n>2. Since there is more

than one agent, communication must take place in between them. For this reason, one class

out of several useful classes for this is the jade.lang.acl.ACLMessage.

1. Create a USE-CASE Uml diagram to show how you are going to set up your

MAS, which agent should be doing what and is responsible for what. Work

iteratively with this one and update it as something maybe changes in your

approach. This paper might give you an indication on how to do a use-case uml

diagram. Bauer and Odell , UML 2.0 and Agents: How to Build Agent-based

Systems, 2005, but do not over-do the modeling part. One example is shown in

figure 2.

2. The data agent in figure 2 should sense its environment and cache streaming data.

Note in this case there is no sensor, the agent should instead be able to read text

files of type *.csv, *.txt. It should act as a simple reflex agent and should be able

to distribute data/information to other agents. In your documentation you must

describe the content of the data. To keep the complexity level in a more

reasonable fashion the data will not be streamed. Use the datafiles from the data

part above in this instruction report.

3. The classification agent should possess knowledge of supervised learning and

more specifically classification by usage of k-NN. Depending on the incoming data

from the other agent and additional information it must be able to be normalized

or standardized before the classification of test and training takes part. The

classification agent need connection to R via Rserve. So the agent get

data/information from the data agent, connects to R and do the performance

measures.

The ones needed for grade 3p is:

? Preform data validation, use K-fold cross validation* and use the results to compute:

o Confusion matrix*

o Accuracy*

o Precision (exactness of a mode)*

(* means explain these in the oral presentation, what they do and how you interpret their results)

And then present/displays the results back in the Java ouput (in netbeans or

eclipse).

To get started you can go back and check the bookseller example, perhaps there is

key concepts there you can reuse for this assignment. Check out where the buyer

agent compares and selects the cheapest book. This might be a place to insert

something more intelligent like classifying data, i.e. contact R via RServe.

Figure 1 Description of the suggested system

Figure 2 Example of agent skills and dependencies

For grade 4p:

All of grade 3p tasks need to be working.

1. One of your agents has already been developed for performing supervised learning by using

the kNN-algorithm i.e. the Classification Agent. Now, develop another new agent who should

possess the ability to perform unsupervised learning /clustering, by making use of k-Means

clustering. The clustering agents' knowledge and abilities are realized by connecting the agent

to a R via Rserve. The data agent should (as before) forward its' processed data, but since

there are now two agents it depends on the content and structure of the data. This new

clustering agent should only be invoked if and only if there is not a label (i.e a target class like

in supervised learning) containing the string "target" as its column name. The agent should

receive data/info. from the data agent. (See the use case UML below)

2. One of the receiving/analyzing agents should be able to produce and display graphical plots.

The agents' ability is realized by connecting the agent to a R via Rserve and use the script in

and as mentioned above the data is provided from the data agent.

3. Do more performance measures then for grade 3 in addition to those also explain and

use/compute the following concepts on your results (they need to be included in the code, the

oral presentation and the report): Recall (completeness) of a mode* & the F1 Score*.

Environment Data

Agent

Data Acquiring

Extract,

Transform, and

Load data

Classification

Agent

Transmits

data

Rserve

Presents analysis

Environment Data

Agent Extract,

Transform

, and

Load data

Transmits

data Classification

Agent

OR

Data Acquiring

Rserve

Clustering

Agent

Presents results of analysis

Rserve

For grade 5p:

All of grade 3p and 4p tasks need to be working.

1. There are a lot of algorithms for supervised classification and regression learning. Using the

same data set as in grade 3p demonstrate the algorithm Naive Bayes for classification (in the

code). Since this is (probably) a new algorithm for you, start with some examples in the

section References (you find literature there). ln your presentation: Display and present the

code and its execution when you run the code and also explain the classification* oral and in

the report.

References

JADE Tutorial and Primer

Rserve Java APl documentation

Naive Bayes Algorithm

Naive Bayes in R example lris Data

Naive Bayes classification in R using the Titanic data set

Random Forest Algorithm

Random Forest - Fun and Easy Machine Learning (8min)

Random Forests in R I Random Forest Classifier (41min)

Random Forest Tutorial I Random Forest in R (1h 7min)

Good luck

Kristin (Titti) Hane khe@du.se


版权所有:留学生编程辅导网 2018 All Rights Reserved 联系方式:QQ:99515681 电子信箱:99515681@qq.com
免责声明:本站部分内容从网络整理而来,只供参考!如有版权问题可联系本站删除。