how to give a research talk - aalborg...

How to give a research talk

Thomas D. Nielsen

September 2008

How to give a research talk September 2008 1 / 23

The purpose of your talk ...

... is not to

impress the audience with your brainpower

tell them all you know about the topic

present all the technical details

Introduction How to give a research talk September 2008 2 / 23

The purpose of your talk ...

... is to

give the audience a taster for your work

present the key ideas, intuitions, and results

make the audience interested

Introduction How to give a research talk September 2008 3 / 23

Preparing and structuring the presentation

Preparing and structuring the talk How to give a research talk September 2008 4 / 23

Preparing the presentation

Deciding on what to say and what to omit

Who is the primary audience?

If only one thing should be remembered from the talk, what should it be (be specific, what isthe key idea)?


Preparing the presentation

Deciding on what to say and what to omit

Who is the primary audience?

If only one thing should be remembered from the talk, what should it be (be specific, what isthe key idea)?

In general

Convey the essential part of your paper, but don’t overwhelm the audience with too much material.

It is better to leave out certain details than to give a superficial treatment of everything or tooverrun your time.

Adopt a nonuniform approach when preparing the presentation.

That does not mean holding back important details - merely omitting less important ones!


Structuring the presentation

Guiding the audience

People often use a contents slide:

Introduction

Contents

Introduction

Bayesian networks

Graph concepts

Domain graphs

Triangulated graphs

Join graphs

Join trees

Junction trees

Message passing

Thomas D. Nielsen

How to give a research talk

Reconsider!Instead

it can be useful to show an outline slide at the start of a section, to help the audience stay ontrack (or help those who got distracted or lost to rejoin you).



Introduction

Remember: If you bore the audience the first few minutes, you may never get them back ⇒ jumpright in!

Give an example to motivate the problem you are working with!Avoid launching into technical details that will confuse the audience.



Introduction


Give an example to motivate the problem you are working with!

Avoid launching into technical details that will confuse the audience.

Conclusion

For conferences, end your presentation with a contributions/conclusions slides to help theaudience remember what to take home from the presentation.

What should be the last thing the audience sees?

Don’t be afraid to include open problems.



Introduction


Give an example to motivate the problem you are working with!

Avoid launching into technical details that will confuse the audience.

The body

. . . What to put in . . .

Conclusion

For conferences, end your presentation with a contributions/conclusions slides to help theaudience remember what to take home from the presentation.

What should be the last thing the audience sees?

Don’t be afraid to include open problems.


What to put in?

?

What to include? How to give a research talk September 2008 8 / 23

What (not) to put in?

It is usually difficult to follow highly abstract presentations!

Definition

1 X: mixed n-dimensional random vector. Y = (Y1, . . . , Yd ), Z = (Z1, . . . , Zc) its discrete andcontinuous parts. A function f : ΩX 7→ R

+0 is a Mixture of Truncated Exponentials potential

(MTE potential) if for each fixed value y ∈ ΩY of the discrete variables Y, the potential overthe continuous variables Z is defined as:

f (z) = a0 +m

X

i=1

ai exp

8

<

:

cX

j=1

b(j)i zj

9

=

;

for all z ∈ ΩZ, where ai , b(j)i are real numbers.

2 f is an MTE potential if there is a partition D1, . . . , Dk of ΩZ into hypercubes and in each Di , fis defined as above.

Ask yourself

Have I illustrated this concept/definition/theorem . . . with an example?



Examples

Use examples to motivate and help the audience

Convey basic intuition

Illustrate main idea

Show extreme cases

. . .



Examples




Show extreme cases

. . .

Triangulation

Triangulation by elimination

1 2

3 4

765

8

Eliminating a node: Make all its noneliminated neighbors complete and remove the node.



Examples




Show extreme cases

. . .

Triangulation


1 2

3 4

765

8 8

1 2

3 4

765




Examples




Show extreme cases

. . .

Triangulation


1 2

3 4

765

8

1 2

3 4

765

8




People can only read or take in very limited information: six or seven things on a slide is quiteenough.

Slides shouldn’t repeat what you plan to say, but rather emphasize it.

Plan to talk about what’s on the slides rather than read them.






Contents

Don’t overwhelm the audience with mathematical details. Remember to explain the(non-standard) notation being used. Maybe include a few backup slides.

Technical details

Q =1

2

NX

i=1

tr (Γ−1yi

E(XXT| Di ) +

NX

i=1

µTyi

Γ−1y1

E(X | Di ) −

|sp(Y )|X

h=1

#yh

2µ

Tyh

Γ−1yh

µyh− . . .

Use descriptive slide titles.

Avoid a presentation that is just dozens of pages of text.

Use figures! But be also sure to explain them.

Use colors (when it is meaningful).






Rule of thumb: about 2–3 minutes pr. slide.


Working out the slides

Style

Do not use non-single-color backgrounds, transition effects, and similar eye candy. It shift thefocus from what is important, the contents.

Avoid jitter when using overlays.



Style



Avoid



Style



Avoid using



Style



Avoid using the



Style



Avoid using the reveal



Style



Avoid using the reveal technique



Style



Avoid using the reveal technique too



Style



Avoid using the reveal technique too excessively



Style



Avoid using the reveal technique too excessively unless



Style



Avoid using the reveal technique too excessively unless there



Style



Avoid using the reveal technique too excessively unless there is



Style



Avoid using the reveal technique too excessively unless there is a



Style



Avoid using the reveal technique too excessively unless there is a point.



Style



Avoid using the reveal technique too excessively unless there is a point.

Use a sans-serif font for your slides.

Handwritten slides are fine, but use permanent ink.


The presentation

The presentation How to give a research talk September 2008 13 / 23

The presentation

Point at the screen not at the laptop

Speak to someone at the back of the room

Make eye contact with the audience (don’t talk to thescreen). This makes them more “involved” and alsohelps you “read” their reactions.

Use rehearsal talks, rehearsal talks, rehearsal talks,. . .


The presentation

Point at the screen not at the laptop

Speak to someone at the back of the room

Make eye contact with the audience (don’t talk to thescreen). This makes them more “involved” and alsohelps you “read” their reactions.

Use rehearsal talks, rehearsal talks, rehearsal talks,. . .

Most importantly: Be enthusiastic!


Looking up material


References

A couple of references

Michael Ernst, Giving a technical presentation (giving a talk), January, 2005 (revised February2008). http://people.csail.mit.edu/mernst/advice/giving-talk.html.

S. L. P. Jones, J. Hughes, and J. Launchbury. How to give a good research talk, ACMSIGPLAN Notices, Vol. 28, No. 11, pp 9–12, November 1993.

I. Parberry. How to present a paper in theoretical computer science: A speaker’s guide tostudents. ACM SIGACT News, Vol. 31, Issue 1, pp. 77–86, March 2000.


http://people.csail.mit.edu/mernst/advice/giving-talk.html

how to give a research talk - aalborg...

Documents