Hybrid Artificial Intelligence System for Cyber Security

Democritus University of Thrace
Dep. of Forestry & Management of the Environment & Natural Resources
Forest Informatics Laboratory
Konstantinos Demertzis – Lazaros Iliadis
ESADM
ECISMD
Hybrid Artificial Intelligence System for Cyber Security

2
 Agenda
– Introduction
– Hybrid Artificial Intelligence System for Cyber Security (HAISCS)
– Evolving Spiking Anomaly Detection Model (ESADM)
– Evolving Computational Intelligence System for Malware Detection (ECISMD)
– ESADM
– Spiking Neural Network Classification
– Spiking Neural Network Pattern Recognition
– ECISMD
– Spiking Neural Network Classification
– Evolving Classification Function (ECF)
– Genetic Algorithm for Offline ECF Optimization
– Results
– Future Directions
– Conclusions

3
 Introduction
– Artificial Intelligence (AI)
– is the intelligence exhibited by machines or software, and the branch of computer
science that develops machines and software with intelligence.
– Machine Learning
– a branch of artificial intelligence, concerns the construction and study of systems
that can learn from data.
– Pattern Recognition
– in machine learning aims to classify data
(patterns) based on either priori knowledge
extracted from the patterns.
– Classification
– is the problem of identifying to which of a set of
categories (sub-populations) a new observation
belongs, on the basis of a training set of data
containing observations (or instances) whose
category membership is known (Supervised
Learning).

4
 Hybrid Artificial Intelligence System for Cyber Security (HAISCS)
Hybrid
Evolving
Spiking
Anomaly
Detection
Model
(HESADM)
ESADM
ECISMD

5
 Evolving Spiking Anomaly Detection Model (ESADM)
ESADM

6

7

8
– Evolving Spiking Neural Network (eSNN) Classification
– Gaussian Receptive Fields
– Rank Order Population Encoding
– One-Pass Learning

9

10

11

12

13
Traf_Red_Full Dataset
Classifier
Train
Accuracy
Test
Accuracy
NaiveBayes 96.387% 95.3981%
RBFNetwork 94.9734% 93.3281%
MLP 97.9475% 97.3743%
LibSVM 98.9691% 97.0335%
k-NN 97.5435% 97.4452%
J48 97.619% 97.4909%
RandomForest 97.57% 97.5046%
LogisticRegression 97.8937% 96.9008%
BayesNet 97.9025% 96.9237%
AdaBoost 96.0311% 95.947%
eSNN 98,9% 97,7%
normalFull Dataset
Classifier
Train
Accuracy
Test
Accuracy
NaiveBayes 99.5112% 98.895%
RBFNetwork 99.9351% 99.4412%
MLP 99.9818% 99.8992%
LibSVM 99.673% 99.1088%
k-NN 99.2554% 98.9278%
J48 99.7751% 99.719%
RandomForest 99.8463% 98.9561%
LogisticRegression 98.998% 98.9855%
BayesNet 98.9933% 98.9718%
AdaBoost 99.2784% 98.9357%
eSNN 99.999% 99.9%

14
 Evolving Computational Intelligence System for Malware Detection (ECISMD)
ESADM
ECISMD

15

16

17
– Evolving Classification Function (ECF)
– used for pattern classification, generates rule nodes in an N dimensional input
space and associate them with classes. Each rule node is defined with its centre,
radius (influence field) and the class it belongs to. A learning mechanism is
designed in such a way that the nodes can be generated.
Rule 1:if
X1 is ( 2: 0.50 )
X2 is ( 1: 0.69 )
X3 is ( 1: 0.95 )
X4 is ( 1: 0.95 )
X5 is ( 1: 0.94 )
X6 is ( 1: 0.52 )
X7 is ( 1: 0.95 )
X8 is ( 2: 0.87 )
X9 is ( 2: 0.82 )
then Class is [1]
Radius = 0.022719 , 20 in node

18

19
 Evolving Computational Intelligence System for Malware Detection (ECISMD).
– Genetic Algorithm for Offline ECF Optimization
– A Genetic Algorithm is an evolutionary
algorithm in which the principles of the
Darwin's theory of evolution are applied to
a population of solutions to a problem in
order to "breed" better solutions.
– Solutions, in this case the parameters of the
ECF network, are encoded in a binary string
and each solution is given a score depending
on how well it performs.
– Good solutions are selected more frequently
for breeding, and are subjected to crossover
and mutation (loosely analogous to those
operations found in biological systems).
– After several generations, the population of
solutions should converge on a "good"
solution.

20

21

22
Packed Dataset
Classifier
Train
Accuracy
Test
Accuracy
RBFNetwork 98.3085% 98.0859%
NaiveBayes 98.3975% 97.1144%
MLP 99.5326% 96.2189%
LibSVM 99.4436% 89.8507%
k-NN 99.4436% 96.6169%
eSNN 99.8% 99.2%

23
Malware Dataset
Classifier
Train
Accuracy
Test
Accuracy
RBFNetwork 94.4031% 93.0612%
NaiveBayes 94.0533% 92.3469%
MLP 97.7551% 97.289%
LibSVM 94.6218% 94.2857%
k-NN 98.1198% 96.8367%
ECF 99.05% 95.561%
Optimized
ECF
99.87% 97.992%

24
– Future Directions

25
 Demertzis K., Iliadis L. (2014). A Hybrid Network Anomaly and Intrusion Detection Approach Based on
Evolving Spiking Neural Network Classification. In: Sideridis A., Kardasiadou Z., Yialouris C., Zorkadis V.
(eds) E-Democracy, Security, Privacy and Trust in a Digital World. e-Democracy 2013. Communications in
Computer and Information Science, vol 441. Springer, Cham
 Demertzis K., Iliadis L. (2014). Evolving Computational Intelligence System for Malware Detection, In:
Advanced Information Systems Engineering Workshops, Lecture Notes in Business Information
Processing, 178, 322-334. doi: 10.1007/978-3-319-07869-4_30
 Demertzis K., Iliadis L. (2014, April). Bio-Inspired Hybrid Artificial Intelligence Framework for Cyber
Security. In: Daras N., Rassias M. (eds) Computation, Cryptography, and Network Security. Springer,
Cham
 Demertzis K., Iliadis L. (2014, November). Bio-Inspired Hybrid Intelligent Method for Detecting Android
Malware, In: Iliadis L., Papazoglou M., Pohl K. (eds) Advanced Information Systems Engineering
Workshops. CAiSE 2014. Lecture Notes in Business Information Processing, vol 178. Springer, Cham
 [Demertzis K., Iliadis L. (2015, April). Evolving Smart URL Filter in a Zone-based Policy Firewall for
Detecting Algorithmically Generated Malicious Domains. In: Gammerman A., Vovk V., Papadopoulos H.
(eds) Statistical Learning and Data Sciences. SLDS 2015. Lecture Notes in Computer Science, vol 9047.
Springer, Cham.
 Demertzis K., Iliadis L. (2015, September). SAME: An Intelligent Anti-Malware Extension for Android ART
Virtual Machine. In: Núñez M., Nguyen N., Camacho D., Trawiński B. (eds) Computational Collective
Intelligence. Lecture Notes in Computer Science, vol 9330. Springer.
 Demertzis K., Iliadis L. (2016), Computational Intelligence Anti-Malware Framework for Android OS,
Vietnam J Comput Sci (2017) 4: 245. https://doi.org/10.1007/s40595-017-0095-3.

26
 Demertzis K., Iliadis L. (2016), Ladon: A Cyber-Threat Bio-Inspired Intelligence Management System,
Journal of Applied Mathematics & Bioinformatics, vol.6, no.3, 2016, 45-64, ISSN: 1792-6602 (print), 1792-
6939 (online), Scienpress Ltd, 2016.
 Demertzis K., L. S. Iliadis, V.-D. Anezakis, An innovative soft computing system for smart energy grids
cybersecurity, Advances in Building Energy Research, pp. 1-22, Taylor & Francis.
 K. Demertzis and L. Iliadis, “The Impact of Climate Change on Biodiversity: The Ecological Consequences
of Invasive Species in Greece”. In: Leal Filho W., Manolas E., Azul A., Azeiteiro U., McGhie H. (eds),
Handbook of Climate Change Communication: vol. 1, Climate Change Management, pp.15-38. Springer,
Cham, 2018. https://doi.org/10.1007/978-3-319-69838-0_2
 K. Demertzis, L. Iliadis and V.D. Anezakis, “A deep spiking machine-hearing system for the case of
invasive fish species”, Proceedings of 2017 IEEE International Conference on Innovations in Intelligent
Systems and Applications, Gdynia, Poland, pp. 23-28, 2017. doi:10.1109/INISTA.2017.8001126
 K. Demertzis, L.S. Iliadis and V.D. Anezakis, “Commentary: Aedes albopictus and Aedes japonicus—two
invasive mosquito species with different temperature niches in Europe”, Frontiers in Environmental
Science, vol.5, no. 85, pp. 1-3, 2017. doi: 10.3389/fenvs.2017.00085
 K. Demertzis and L. Iliadis, “Detecting invasive species with a bio-inspired semisupervised
neurocomputing approach: the case of Lagocephalus sceleratus”, Neural Computing & Applications,
vol.28, no.6, pp. 1225-1234, 2017. Springer, London. https://doi.org/10.1007/s00521-016-2591-2

27
 Forest Informatics Laboratory
 Director of the Lab Professor Lazaros S. Iliadis
 Research Areas
– Fuzzy Logic
– Computational Intelligence
– Soft Computing
– Machine Learning
– Pattern Recognition
– Neural Networks
– Support Vector Machines
– Genetic Algorithms
– Adaptive Fuzzy Clustering
– Heuristic Models
– Intelligent Agents – multiAgent Systems
– Expert Systems - Knowledge Systems - Fuzzy Inference Systems
– Intelligent Information Systems and Applications in Risk Management

28
kdemertz@fmenr.duth.gr | liliadis@fmenr.duth.gr
 Conclusion

Hybrid Artificial Intelligence System for Cyber Security

More Related Content

Hybrid Artificial Intelligence System for Cyber Security