Example Exam - Solutions

7/31/2019 Example Exam - Solutions

1/22

Example Exam Questions and

SolutionsResearch Methods

MAR8001

Richard BirminghamJanuary 2010


2/22

====================

Q1. Discuss the reasons that research isundertaken, presenting this graphically ifyou prefer. Explain what is meant by thespectrum of types of research activity,

and indicate who might be thebeneficiaries of the different types of

activity.


3/22

The Research Rational

A ProblemMoney

Research

New Knowledge

Esteem/Reputation

Higher degree

Personal satisfaction

etc.

Increased profitsImproved planning

etc.

Enhancedquality of life

etc.


4/22

The Research Spectrum

The extreme types of research are:

Blue sky - no obvious benefits, simplyexploring an area of interest to gain new

knowledge

Near market - immediate commercial benefitexpected

In practice research can be anywherebetween these two extremes


5/22

Research Beneficiaries

Who will benefit influences type ofresearch and partnership:

Benefit to researchers from scholarships,fellowships, research clubs

Benefits to society from blue sky research

Benefits to funding body from industrial R & D

Distinctions can be blurred


6/22

==================

Q2. Draw a diagram of the interactions betweenfunding bodies and research organisations,indicating how a research organisation responds

to a call for proposals. Describe how to plan forsubmitting a Research Proposal in response tosuch a call, and indicate what should beconsidered in preparing the document, and what

is likely to be discussed in negotiations aftersubmission.


7/22

Standard Interactions

Problem Identification

Call for Proposals Project Proposal

Evaluation Negotiation andContract Agreement

RESEARCH

Reports

Final ReportDeliverables

Project

Meetings

Acceptance

Exploitation Dissemination

FUNDERRESEARCHER

(stage payments)

(final payment)


8/22

Planning the Proposal

Write proposal for a specific sponsor

Obtain background information abouttargeted sponsors policy (from Annual

report, Handbooks or Web site)

Establish if there are deadlines to meet

Establish if there is a particular formatrequired, or a standard application form


9/22

Document Style

Be precise in obeying all instruction (dont

be disqualified from funding for trivialreasons)

Try to avoid using technical languagewhich a non-specialist not understand

Usually addressing two audiences:

Specialist academic referees

Non-specialists on the decision making panel


10/22

Contract Negotiations

Once the proposal is acceptednegotiations will be necessary

These will involve direct meetings withthe funders, and possibly aresubmission of the proposal

Details to agree will include:

Workpackage content and deliverables

Payment schedules

Progress meetings and reports


11/22

======================

Q3. Explain why careful monitoring of aresearch project is necessary. Meetingsare an important part of the monitoring

process. Describe the environment withinin which such meetings take place andhow the decisions are recorded. Discuss

the concept of an informal agenda atsuch meetings.


12/22

Objective of Monitoring

To review progress on the project:

demonstrating that the intended work isprogressing satisfactorily

or re-schedule and re-focus project whendifficulties have been encountered

(approval should be obtained BEFORE

changing work programme) To demonstrate that allocated funds are

being well spent


13/22

Environment of Reporting

Written Reports and Meetings are scheduled inProject Proposal, so maintain schedule

Normally representatives of funding body are welldisposed toward researchers

sympathetic to difficulties encountered need to justify expenditure to their superiors

Changing track without seeking approval createsdifficulties

Calls for further funding creates difficulties


14/22

Meeting Notes

Large projects, with many participants, willhave secretary taking minutes

Small projects: after meeting write upMeeting Notes and distribute to all

Notes outline the meeting, in particulardetailing:

agreed actions

changes to Project Plan


15/22

Informal Agenda

The meeting is an important opportunity to exploreideas that may not be included on the formal agenda,for example:

difficulties encountered (change of staff, internal

politics, other funders/researchers) warn of difficulties anticipated

discuss further research, and further funding?

related projects

Cultivate a mutually beneficial relationship


16/22

=====================

Q4. How are the results of a researchproject normally delivered to the fundingbody? Describe other ways that the results

can be disseminated, detailing thedifferences between primary andsecondary dissemination routes.


17/22

Dissemination Routes

Required reports:

Final report for funders

Degree dissertations and theses

Publications:

Papers in Journals (academic or professional)

Conference Papers

Books

Articles in trade journals or newspapers TV or radio appearances


18/22

Primary Dissemination Routes

Journal Papers: Peer review ensures that these are high quality

and therefore regarded most highly, but in realityreadership may be very limited

Conference Papers: Easier to publish (organisers want delegates so

accept papers readily) so considered of lowervalue, but can target a specific audience

Books: in some disciplines (not engineering) booksare the accepted route


19/22

Secondary Routes

Primary routes have very small readership

Brief summaries of results can reach a much wideraudience through secondary routes

Advertise results to a larger audience, some maythen read the Journal paper:

trade journals reach practitioners

newspapers reach the general public

TV/Radio: huge audience for brief soundbite


20/22

===================

Q5 . What are the main limitations inapplication of ANN methodology?


21/22

Solution: Artificial neural networks, although very promising in many fields

and have advantageous in relation to traditional computingtechniques, suffer from certain limitations. These problems could beexternal to the network, for example limitation in hardwaretechnology or could be internal such as deficiencies in training

algorithms etc. Some of the most significant problems could be: Long training time. Quick convergence is not guaranteed in any

training algorithm. Generally there will be a quick drop in totaltraining error at the beginning of the training, but as the training goeson the error reduction rate will get smaller.

Network structure for new problems must be selected by trial and

error. Although many attempts have been done, there is no clearprescription on how to choose the network topology for a successfulgeneralisation or pattern classification.


22/22

Any prioriknowledge about the data cannot be transferred to thenetwork parameters. Although that knowledge may help in theselection of network structure.

In many engineering applications, network does not provide anyphysical sense of its finding in relationships. All the data, no matterwhat they are, are treated the same, however their influence on the

network output will inherently be taken into account. Functional relationship provided by a trained network by no means

is similar to the known and well-established mathematical models,such as regression techniques.

During training, the weights can be adjusted to very large values,causing the total input of a hidden or output unit to reach very high

positive or negative values. Sigmoidal activation function will eithersaturate to one or zero respectively, meaning that the neurone doesnot respond or sense any more changes, a situation callednetwork paralysis.

Example Exam - Solutions

Documents