Transcript Mathematics and Cybercrime

Mathematics and Cybercrime
Dr. Robert Layton
Internet Commerce Security Laboratory
Centre for Informatics and Applied Optimisation
University of Ballarat
1/24
Who am I?


Bachelor of Computing, Major in Mathematics (2007)
Bachelor of Applied Computing
(first class honours) (2008)

UB Council – Student Member (2009)

PhD (2011), ICSL

Postdoc, ICSL

Early Career Researcher
2/24
Overview
- What I learnt during my Bachelor.
- How I learnt it.
- What worked, what did not.
- How I apply that today in my research.
3/24
Early On
- Always good at Maths.
- Natural to continue that in Uni (good grades,
less essay writing).
- All electives were Maths related, ran out of
subjects to do, so started doing guided studies.
4/24
Game Theory
Early unit on game
theory.
Introduction to game
state trees.
Example based
teaching, able to see
the outcomes quickly.
This comes in handy
later.
5/24
Modelling Reality
- Differential Equations
- A significant leap in
difficulty for me.
- Good communication
from lecturer helped
significantly.
- High Distinction for
subject, giving a huge
confidence boost for
future subjects.
6/24
Prabhu Manyem
Linear Programming
Another difficult subject,
but introduced
practical applications
of matrices, which
have become
incredibly important
later on.
7/24
Automata and Context-Free
Languages
8/24
Combinatorial Optimisation
- A branch and
bound approach
to integrated
circuit design.
- NP-hard problem,
related to the 2dimensional Bin
Packing problem.
- Simulation 9/24
Topology
Guided study with David
Yost, using the book
“Topology without
Tears” by our ex-head
of school, Sid Morris.
Defined the concept of
metric spaces for me,
which helps later.
10/24
Artificial Intelligence
- Implemented a
program for playing
Reversi/Othello
- Searching the game
state tree using the
minimax algorithm
with alpha-beta
pruning
- Depth first search for
memory conservation
11/24
Neural Networks
- Taught by Richard
Dazeley.
- Heavy experimental
approach – code it
and see how it works!
- Fitted my learning
style really well.
- Introduction to Data
Mining
12/24
Data Mining
Applied a data mining
approach to poker.
Then went with an
artificial intelligence
approach based on
statistical stereotyping
of opponent play.
Honours Project
13/24
My thoughts on my
undergraduate education.
Interest Matters
1) Getting students to understand why, rather than
what, helps.
2) Goal-based work - “Develop a program that
uses minimax to play a computer game”
3) Get people's hands dirty – examples are
needed, not only proofs but applications.
14/24
My Research – in numbers
>$1,088,000 in successful grants
- NeCTAR - >$1M (still confidential)
- AFACT - $25,000 + $15,000
- Google - $15,000
- APWG - $3000
- Other - $30,000
(distributions matter!)
20 publications
- 4 in Journal of Natural Language
Engineering
- 2 in other journals
- 14 conference papers
- 15 co-authors across five institutions
15/24
Supervision – complete: 2
- 1 honours
- 1 undergraduate guided study
Supervision – in progress: 7
- 1 full time PhD
- 4 part time PhD
- 1 full time Masters by Research
- 1 undergraduate guided study
My Research Goal
Automated
–
From start to finish
Unsupervised
–
No training set
Authorship
–
Who wrote what?
Analysis
–
How does that help us?
16/24
Impact – Phishing Attacks
17/24
18/24
Educational Underpinnings
Without a solid background in Mathematics,
my research findings to date would not
have been as effective.
19/24
Statistics: Distributions Matter
20/24
Topology: Distance Metrics
21/24
Modelling: Temporal Analysis
22/24
More on Temporal Analysis
23/24
Lessons I learnt
- Blind Experimentation will only get you so far
- Understanding the application, rather than “black
boxing” it will net better results.
- Learning how algorithms work is more efficient in
the long run.
- Learn widely – concepts from other fields will
help where you least expect it.
24/24