Applied Computing is a field within SCIENCE which applies practical approaches of computer science to real world problems

1.

Prof. Dhiya Al-Jumeily
Professor of Software Engineering/Associate Dean
Faculty of Engineering and Technology
[email protected]

2.

3.

WHERE IS LIVERPOOL
• Good transport links with the rest of Britain
• 40 minutes from Manchester International Airport
• 2 hours by train from London
• One of the UK’s top tourist destinations
• Home for Liverpool and Everton football clubs

4.

Liverpool is officially the UK’s friendliest City
(Condé Nast Readers Travellers Awards)

5.

6.

7.

8.

9.

10.

11.

12.

13.

14. COMPUTING

Inspired By The Past
Motivating The Future

15.

Conventional Vs Biologically Inspired Computing
Big Data and Data Science
Application of Data Science in Healthcare and Medicine

16. What is computing?

• Applied Computing is a field within SCIENCE
which applies practical approaches of
computer science to real world problems
across multiple disciplines to produce
effective solutions.
• The techniques used within computing open
up possibilities for other disciplines which can
be explored for solutions and improvements
to tricky problems which ultimately are either
too complex or big for humans to provide the
answers.
• Applied computing uses various computer
science techniques such as algorithms,
databases, networks and more modern
applications such as deep learning, Big Data
and more to build systems and applications
which can be used in the real world.
• Applied computing has been effective
in many areas for developing solutions
which can analyse data to provide
informative results for applications in
real-world situations, the following
branches are some of those in which
applied
computing
has
been
effectively used:
Education and Learning Techniques
Health and Medicine
Gaming
Business
Economics

17. Progress in Early Computing

• Applications in computing and information
processing have existed since ancient times.
Such precursors of the modern computing era
now serve to demonstrate both the
conceptual and technological difficulties in the
evolution of such processing capabilities
during the pre-digital age.
• Early examples of applied computing include
the well known abacus, the earliest
computational instrument on record, along
with less widely known mechanical artefacts
such as the Antikythera mechanism, an
intricate analog device developed during the
first century BC to calculate astronomical
events.

18. Progress in Early Computing

• The less widely known mechanical artefacts such
as the Antikythera mechanism, an intricate
analog device developed during the first century
BC to calculate astronomical events.
• However, despite the endurance of both manual
arithmetic devices and one-of-a-kind automatic
mechanisms during periods of antiquity, the
position of computing remained largely
unchanged up until 1800, a time at which the
effects of the industrial revolution had become
well established. It is estimated that fewer than
100 automatic computing mechanisms were ever
built prior to this time.

19. Mechanical Computing and the Industrial Revolution

• Following the turn of the 19th Century,
a sustained transition began to occur
from devices that could perform
arithmetic, to devices that could
represent logic through mechanical
operation.
During
this
time,
commercially viable devices that could
accept task-specific "programs” were
developed, such as the Jacquard loom
developed in 1804.

20. Plans for Babbage’s Analytical Engine

• Charles Babbage, thought to be
inspired by such inventions,
developed
designs
for
two
automatic devices, the "Difference
Engine" and the "Analytical
Engine". The latter represented a
conceptual breakthrough, a sketch
for the first programmable digital
computer. Though conceptually
sound, the Analytical Engine was
never realised due to the cost and
difficult demanded by such an
undertaking at the time.

21. Introduction of Conventional Computing

• Automated applied computing began in the form of machines that were
created to perform information processing tasks, specific to a goal of
interest. The physical machine could be programmed to alter the nature
of the task, for example in the case of the Jacquard loom. However, if the
goal of interest was to change, the physical machine itself, the substrate
upon which the computation was run, would have to be redesigned
and reconstructed. In 1936, a theoretical concept known as the
"Universal Turing Machine" was introduced by Alan Turing in a paper
titled "On computable numbers, with an application to the
Entscheidungsproblem".
• Turing had discovered a property of computation that would enable the
full potential of computing applications to be unleashed, he found that
with the provision of a set of simple primitive operations, the space of all
possible computations could be represented on any machine without
undertaking a redesign of the machine itself. Conversely, the same
realisation simultaneously proved that most previously developed
machinery was fundamentally limited, since the not all of the so called
primitives were incorporated into these devices. The notion of "Turing
Completeness" is now well recognised and is fundamental to the design
of all general purpose computing systems, having opened the way for
both modern applied computing research and the exploration of artificial
intelligence.

22.

Conventional Vs Biologically Inspired
Computing

23. Conventional Computing

A computer does what it is programmed to do !
Conventional computers follow a set of ‘well-defined’
instructions in order to solve a problem.
When equipped with the appropriate rules, it can
perform many tasks better or faster than humans.
playbuzz.com
BUT
conventional computing ‘struggles’ with certain
types of tasks !
Computer performed better than the human at chess in
the late 1990’s !

24.

25.

26.

Aoccdrnig to a rscheearch at an Elingsh uinervtisy, it deosn't mttaer
in waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is
taht frist and lsat ltteer is at the rghit pclae. The rset can be a toatl
mses and you can sitll raed it wouthit porbelm.
[rscheearch at Cmabrigde Uinervtisy]

27.

biometricupdate.com/
inhabitat.com

28. Conventional Computing - Challenges

Examples:
Scenarios associated with
Recognising complex patterns such as human
faces or emotions
Limited or no knowledge
Recognising a variety of handwriting.
Context
Identifying the words and context in spoken
language.
Driving a car through busy streets.
New or previously unseen situations
Decision making
Extremely large number of possible solutions to
choose from

29.

Biologically Inspired Computing

30. Biologically Inspired Computing

Scientists and engineers have been learning a great deal by
studying natural and biological systems and modelling some of
their problem solving and adaptation strategies
Bio-inspired computing:
Computing based on ideas, principles and concepts from
biology and nature

31. Biologically Inspired Computing

Biological sources of inspiration
Human Brain
Natural evolution
Swarms:
Ants
Bees / Wasps
Immune System
DNA
Birds / Fish
Others !
Algorithm / Computing paradigms
Artificial Neural Networks (Rosenbl’t 1958, Rumelhart 1986).
Genetic Algorithms (Goldberg and David, 1989)
Particle Swarm Optimisation (Kennedy, 1995)
Ant Colony Algorithms (Dorigo and Di Caro, 1999)
Artificial Bee Colony algorithm (Karaboga, 2005)
Artificial Immune Systems (De Castro, Timmis 2002)
DNA Computing (Adleman 1994)
Flocking Behaviour (Reynolds 1987)
Others !

32. Biologically inspired Computing

Key application domains
Healthcare and Medicine (e.g. image analysis)
Robotics (e.g. path planning)
Security (e.g. intrusion detection)
Speech and Natural Language Processing
ionicon.com
Bio-inspired Computing examples
Neural Computing (Neural Networks)
Evolutionary Computing (Genetic Algorithms)
extremetech.com

33. Neural Computing

Biological Neuron
Inspired by the way the human brain works.
Brains do computation differently from conventional
computers
Artificial Neuron
Activation
function
Learn from examples and experience
Massive parallelism (1010 Neurons)
Neurons communicate with each other through signals
that can be excitatory (+) or inhibitory (-).
The artificial (binary) neuron computes a weighted sum of
the inputs and outputs a value according to whether the
sum is above or below a threshold.
theodysseyonline.com

34. Artificial Neural Networks

A large number of highly interconnected processing
units (neurons) working in parallel to solve a specific
problem.
Not programmed to perform a specific task.
Instead, neural networks learn by example.
Go through a learning (training) process to become an
‘expert’ in a specific task.
Can then be used to perform this task in similar and new
situations or scenarios.
Number of
bedrooms
Age
Size
Postcode
Condition
House
price

35. Financial Time Series Analysis with Neural Networks

2 .5
2 .5
2
2
1 .5
1 .5
1
1
RDP + 5 Values
RDP + 5 Values
Financial Time Series Analysis with Neural Networks
0 .5
0
-0 .5
0 .5
0
-0 .5
-1
-1
-1 .5
-1 .5
-2
-2
-2 .5
-2 .5
0
20
40
60
80
Day
US/EU currency exchange rate
100
0
20
40
60
80
Day
JP/UK currency exchange rate
100

36. Artificial Neural Networks

Key strengths
0.6
0.5
Ability to learn / Adaptation
Fault tolerance
Distributed computing
0.4
0.3
0.2
Key challenges
0.1
Over-learning and under-learning
Representative-ness of the training set
Local ‘optimum’ solutions
0
0
50
100
150
200
250
300
Earthquake/After-shocks with Polynomial Neural
Networks (Tawfik 1998)

37.

Genetic Algorithms (GAs)
GAs adopt mechanisms of natural selection and genetics in
their search for ‘optimum’ solutions.
aspenpublicradio.org/
An optimisation problem is one which has a set of possible
solutions, and the goal is to find ‘best’ solution.
E.g: Traveling Salesman Problem
Cannot evaluate all possible solutions !
computationally expensive, impractical or impossible.
Traveling salesman problem
Given a set of cities and the distances
between them, find the shortest possible
route that visits each city once and returns
to the origin city

38. Genetic Algorithms (GAs)

Evolve a population of
(low-fitness) solutions to a
population of high-fitness
solutions through many
generations.
Use a ‘fitness’ measure to
evaluate the fitness of the
solutions.
ewh.ieee.org

39. Genetic Algorithms (GAs)

Evolve a population of
(low-fitness) solutions to a
population of high-fitness
solutions through many
generations.
Use a ‘fitness’ measure to
evaluate the fitness of the
solutions.
ewh.ieee.org
Selection/reproduction
Give preference to better
solutions (Survival of the
fittest principle).

40. Genetic Algorithms (GAs)

Evolve a population of
(low-fitness) solutions to a
population of high-fitness
solutions through many
generations.
Use a ‘fitness’ measure to
evaluate the fitness of the
solutions.
abrandao.com
Crossover: combination of parents
materials to form children
ewh.ieee.org
Selection/reproduction
Give preference to better
solutions (Survival of the
fittest principle).

41. Genetic Algorithms (GAs)

Mutation:
change to part of
some solutions
Genetic Algorithms (GAs)
Evolve a population of
(low-fitness) solutions to a
population of high-fitness
solutions through many
generations.
Use a ‘fitness’ measure to
evaluate the fitness of the
solutions.
abrandao.com
Crossover: combination of parents
materials to form children
ewh.ieee.org
Selection/reproduction
Give preference to better
solutions (Survival of the
fittest principle).

42. Genetic Algorithms for tactical driving decision making

zipcar.com
GA selects a ‘tactical’ driving decision in terms of change of lane,
change of speed, change of acceleration, etc
Tawfik, H. and Liatsis, P. (2008). An intelligent systems framework for prototyping tactical driving decisions. Intelligent Systems Technologies
and Applications.
Tawfik, H. and Liatsis, P. (2006). Modelling tactical driving manoeuvres with GA-INTACT. Computational Science.

43. Genetic Algorithms

Key strengths
Near optimum solutions
Evolution and adaptability
Distributed computing
Key challenge
Pre-mature convergence to ‘sub-optimal’ solutions

44. Deep Learning Networks (Aizenberg et.al 2000, Hinton 2007)

Multiple layers of learning (internal layers provide learned representations of the input data)
Can be trained on very large collections of images, videos, and speech samples.
Major improvements in recent years in image, speech and handwriting recognition.
Automatic generation of text captions for images with deep networks
In: Jordan, M.I. and Mitchell, T. M. SCIENCE, 2015

45.

Big Data and Data Science

46. What is Big Data?

• Isn’t all data big?
• Big Data: information that can’t be processed or analysed using
traditional processes or tools
• Organisations produce vast amounts of data each and every day
• Some of this is unstructured or at best semi-structured
46

47. Structured and unstructured data

Structured data
• Is laid out according to
some agreed structure or
pattern
• Is easy to find what you
are looking for as it will
always be in the correct
place
• Can be processed by
humans relatively easily
• Think of a phone book:
• Ordered by Surname
• Follows a set pattern for all
entries
Unstructured data
• Has no meaningful pattern
• Think of a wall with space
to write on and a million
people with a marker pen!
• Difficult to find anything
without trawling through
all the data and trying to
make sense of it (not for
humans!)
• Computers can be used to
help process the data and
find things
47

48. Characteristics of Big Data

IBM characterises Big Data by the four Vs
• Variety – the many different
forms that data can take
• Velocity - the speed at which
the data is produced or
needs to be processed
• Volume – the amount of
data produced or consumed
• Veracity – the truthfulness
(or uncertainty) of data
48

49. Big Data science

Adopts the philosophy that with sufficient data,
science can be carried out by directly analysing and
learning from data.
A lot of knowledge discovery can be driven by the
availability of the massive amount of data.
It can be much easier to train a system by
presenting it with examples of desired response
than to program it to predict the response for all
possible scenarios.
Big Data
Big Data
Science
Insights
Intelligence
Knowledge
Descriptive Analytics (What has happened?)
Predictive Analytics (What could happen ?)
Prescriptive Analytics (What should we do?)

50. Data Science: Why all the Excitement?

e.g.,
Google Flu Trends:
Detecting outbreaks
two weeks ahead
of CDC data
New models are estimating
which cities are most at risk
for spread of the Ebola virus.
50

51. Data Science – One Definition

52. Contrast: Databases

Databases
Data Science
Data Value
“Precious”
“Cheap”
Data Volume
Modest
Massive
Examples
Bank records,
Personnel records,
Census,
Medical records
Online clicks,
GPS logs,
Tweets,
Building sensor readings
Priorities
Consistency,
Error recovery,
Auditability
Speed,
Availability,
Query richness
Structured
Strongly (Schema)
Weakly or none (Text)
Properties
Transactions, ACID*
CAP* theorem (2/3),
eventual consistency
Realizations
SQL
NoSQL:
Riak, Memcached,
Apache River,
MongoDB, CouchDB,
Hbase, Cassandra,…
ACID = Atomicity, Consistency, Isolation and Durability
CAP = Consistency, Availability, Partition Tolerance

53. Contrast: Databases

Databases
Data Science
Querying the past
Querying the future
Business intelligence (BI) is the transformation of raw data into meaningful and
useful information for business analysis purposes. BI can handle enormous
amounts of unstructured data to help identify, develop and otherwise create new
strategic business opportunities - Wikipedia

54. Contrast: Scientific Computing

Image
General purpose classifier
Supernova
Not
Nugent group / C3 LBL
Scientific Modeling
Data-Driven Approach
Physics-based models
General inference engine replaces model
Problem-Structured
Structure not related to problem
Mostly deterministic, precise
Statistical models handle true randomness,
and unmodeled complexity.
Run on Supercomputer or
High-end Computing Cluster
Run on cheaper computer Clusters (EC2)

55. Contrast: Computational Science

CASP: A Worldwide, Biannual
Protein Folding Contest
Brain Mapping: Allen Institute,
White House, Berkeley
Quark
Raptor-X
Rich, Complex
Energy Models
Data-intensive,
general ML models
Techniques (Massive ML)
Faithful, Physical
Simulation
Feature-based inference
Principal Component Analysis
Conditional Neural Fields
Independent Component Analysis
Sparse Coding
Spatial (Image) Filtering

56. Contrast: Machine Learning

Machine Learning
Data Science
Develop new (individual) models
Explore many models, build and
tune hybrids
Understand empirical properties
of models
Prove mathematical properties of
models
Improve/validate on a few,
relatively clean, small datasets
Publish a paper
Develop/use tools that can
handle massive datasets
Take action!

57.

Application of Data Science in
Healthcare and Medicine
(Bio-inspired Computing for Big Data Science)

58. Bio-inspired Computing for Big Data Science

Big data approaches need to exploit computing paradigms
that are powerful, fault tolerant and capable of adapting to
the challenging nature of the data.
Scientists and engineers are turning to biologically and
nature-inspired computing, and other AI, techniques to
obtain useful insights from big data.

59. Medical and Health Real World Applications

• In terms of the research proposed
within the previous slides; the
application of applied computing
becomes apparent further to the
end product.
• The research involved within our area is
predominantly theory-based; the applications
of the derived theory can then be used to
provide applied computing opportunities.
• By using Machine Learning (ML) techniques and
statistical methods to classify and produce
predictive outcomes; these conclusive
algorithms and results can be used within
structures such as Decision Support Systems
(DSS) which can aid health professionals in
making decisions about a patients condition,
their treatment options or in outlining the risk
factors associated with a patients’ condition
given their biological make-up or environment.

60. Machine Learning (ML) and its’ applications

WHAT IS ML?
MACHINE LEARNING WITHIN CLINICAL DECISION
SUPPORT SYSTEMS (CDSS)
• The concept of machine
learning refers to a computer
program that able to learn
and gain knowledge from
past experiences and/or
through
identifying
the
important features of a given
dataset in order to make
predictions about other data
that were not a part of the
original training set.
ML considered to be the backbone for the majority of
sophisticated CDSS.
It is one of the principal components of the
information architecture of CDSS.
It is essential part of CDSS and enables such systems
to learn over time.
It would handle more complicated decisions that
might require a specialist knowledge as well as
evaluating the consequences of the suggested
solution.

61. Case Study WHAT IS SICKLE CELL DISEASE (scd)?

Sickle cell disease (SCD) is one of the most prevalent
diseases, which could have an influence on patients’ lives
due to red blood cell (RBC) abnormality.
Life expectancy by 20 to 30 years and affects a large
number of inhabitants worldwide, particularly in Africa
and Asia continent.
SCD can easily inherited to the child throughout genetic
of sickle haemoglobin (Hb S) either from both parents or
from one of them measured as abnormal haemoglobin.
According to the World health Organisation (WHO), 7
million born each year suffers from this disease.
5% of the population around the globe carries traits genes
for haemoglobin disorders.

62.

TREATMENT CHALLENGES
• Continuous self-care monitoring of chronic diseases and
medicine intake are vital for patients to mitigate the severe
of disease by taking the appropriate medicine at proper
time.
• Within this context, there is a significant need for
constructing cooperative care environment to improve
quality of care and increase caregivers’ efficiency with the
purpose of providing regular information for medical
experts and patients.
• In order to achieve that facilities, this research focus on
how to develop an intelligent system to provide short term
interpretation of overall goals in Sickle Cell Disease (SCD).

63. The current situation in healthcare environment

• Currently, all hospitals and healthcare
sectors are using manual approach
that depends completely on patient
input, which can be slowly analysed,
time consuming and stressful as well.
• The most challenging aspects that
facing healthcare in these days is that
there
is
still
insufficient
communication between the SCD
patients and association healthcare
providers.
• There still need for developing of an
intelligent SCD diagnosis system that
eligible to provide a specific treatment
plan inspired by expert system.
• There are still a number of barriers to
obtain excellent communication between
both patient-medical experts relationship
in terms of workload, patients' fear and
anxiety, and fear of verbal or abuse
physical.

64. Genetics

• Branch of science focus in the study of genes, heredity and variation in living organisms.
• DNA
• Genes are made up of DNA. DNA is
essentially a code which uses 4
nucleotide bases to define a ‘message’,
these 4 nucleotide bases are: Adenine
(A), Guanine (G), Cytonsine (C) and
Thymine (T).
TRANSCRIPTION PROCESS
By unravelling the Double-Helix strand, a copy of
the template strand of DNA will be made which
forms what we call ‘RNA’. The Thymine which
was originally in the DNA changes to Uracil
during this process.
• This DNA is then unwrapped from its’
double-helix form and subjected to
various processes in order to produce
RNA through the transcription process
and Proteins from the translation
process. This whole process is called
the ‘Central Dogma’.
TRANSLATION PROCESS
The RNA produced from the transcription will be
used to produce protein in the translation
process. The RNA will be up of a single strand of
nucleotides which will then be separated into
block of 3–
• Within each cell the genetic
information flows from DNA to RNA to
protein.
acg ugg cau ccu gua
These block of 3, ‘codons’, will be translated into
proteins e.g. a c g Threonine

65. Genomics

• WHAT IS IT?
• Genomics is the study of a person’s
complete DNA sequence - including
genes (exons) and “noncoding” (intron)
DNA segments in the chromosomes and
how those genes interact with each
other, as well as the internal and
external environments they are exposed
to.
• WHAT HAS BEEN FOUND?
• Overall it has been estimated that there are
only approximately 24,000 genes in the
human body (previously thought to be
100,000 genes!)
• WHAT HAS BEEN DONE IN THE AREA?
• Human Genome Project is an example
of a considerable achievement in the
area of genomics. It was an international
research effort to determine the
sequence of the human genome and
identify the genes that it contains.
The aim is to find out, at what stage the problems start in the
genome for each disease, disorder and illness; and if they can
find this (can they produce a treatment or prevention?)

66. Bioinformatics: Data Types

Bioinformatics is the application of collecting
and analysing complex biological data such as
human genes.
SINGLE NUCLEOTIDE
POLYMORPHISMS(SNPs)
There are many data types which can be used
to determine different variations and
causations for diseases and disorders.
Types of data we are considering in our
research:
PROTEIN EXPRESSION
RAW DATA
Signal data
CLINICAL DATA
Translated signal data
acgtaatgctattgctccagt

67. Personalised Medicine

Personalised medicine intends for
medication to be tailored to each
individuals body based on their own
personal information, and as a result
the treatment of a patient should be
more
comfortable
with
the
elimination, or at a very minimum
considerable reductions of side effects.
Current and Common Approach: ‘One
size fits all’
Personalised medicine: ‘Catered
Medication’

68. Pharmacogenomics

PHARMACOGENOMICS investigates the impact of
genetic variation(SNPs) or specific genotypes (DNA
sequence) of individuals on their drug metabolism or
drug response.
Each patient reacts differently to the medication;
pharmacogenomics can help to determine whether a
patients’ reaction to medication puts them in the
category of responders or non-responders or results
in bad side effects (Adverse Drug Reaction [ADR]).
Pharmacogenomics can advance the development of
personalized medicine.
PHARMACODYNAMICS
This area focuses on the resulting function of the
medication and the effects . This study takes into
account the drug concentration, site of action and
the resulting effect of the administered drug.
PHARMACOKINETICS
How the body interacts with the drug (the
transportation). This area focuses on how long it
takes for the drug to reach its’ destination (or if it
even will). It measures the ‘time course’ of:
• Absorption
• Distribution
• Metabolism
• Excretion (Elimination)

69. Conclusion

Biologically inspired computing systems ability to adapt, learn/evolve, and their
tolerance to faults and uncertainties, make them a valuable data science tool in the
big data era.
Novel developments/areas of research include:
More sophisticated bio-inspired systems
Collaborative and team based learning
Crowd computing
Quantum computing