A B M Shawkat Ali CQUniversity, Australia

A B M Shawkat AliCQUniversity, Australia

Support Vector Machine in Smart Grid and Cloud Computing

• Smart Grid: How much Computational Intelligence (CI) is involved?

• Cloud Computing: How can CI ensure the services?

• Current Projects

• Demand and Supply• IT• Automatic Observation

Figure 1. A sketch of Smart Grid.

Need an Intelligent System for:

• Forecasting demand and supply• Grid security• Monitoring power quality• Power storage

Let us consider n data points

For instance

Recently, a new loss function called -insensitive loss has been proposed by Vapnik (1995):

Subject to

This optimization problem can be transformed into the dual problem (Vapnik, 1995), and its solution is given by

with coefficient values in the range and denotes the dot product in the input space.

,

Figure 2. Hourly average solar radiation of Rockhampton, Australia

Average solar radiation w/m^2

Time in Hour

Outliers, High-leverage Points and Influential Observations

Figure 3. Outlier mapping.

Imon (2005) defines generalized Studentized residuals and generalized weights (leverage) as

Imon (1996) also introduces generalized potentials for identifying multiple high-leverage points by using group-deletion idea for a dataset as

He re-expresses GDFFITS in terms of deletion residuals and leverages as

Outlier Detection in Linear Regression

Table 1. Solar radiation prediction performance.

• Software• Service• Storage• Data Mining• Data analysis

Figure 4. An overview of Cloud Computing.

Figure 5. A real life Cloud Computing Environment.

Figure 6. Performance chart of Hypervisor at the time of Installing new VM.

Training Data Domain

Non linear Mapping by Kernel

To Choose Optimal Hyperplane

Linear Feature Space of SVM

Figure 7. SVM training process.

Construct Model through Feature

knowledge

Class I

Class II

Test Data Domain

Kernel Mapping

Figure 8. SVM model testing process.

Mercer’s Condition

Figure 7. Ten (10) fold cross validation process.

Figure 8. Attack classification performance in the real life Cloud scenario.

Student Projects:PhD Students Mohammed Mizanul Mazid Topic: Intrusion Detection Using Machine LearningGazi Mohammad Shafiullah Topic: Experimental Investigation and Development of Renewable Energy Integration into the Smart GridMd Rahat Hossain Topic: Hybrid Forecasting System of Renewable Energy with Smart Grid for a Sustainable FutureMohammed Arif Topic: Storage and Its Strategic Impacts on Smart GridMD Tanzim Khorshed Topic: Combating Cyber Attacks in Cloud Computing Using Machine Learning TechniquesMD Akhlaqur RahmanTopic: Data Mining in Telecommunication Industry of Call Records, Customer Profiles and Network DataMaster’s StudentChoudhury Wahid Topic: Cancer Classification by Support Vector Machine using Microarray Gene Expression Data

Personal Projects

• Livestock tacking• Road load estimation for a better plan• Industry automation: Magnesia sorting• Cool train monitoring

High Lime Core

EFH1

EFH2

Analysing : Percent correctConfidence : 0.05 (two tailed)Date : 7/17/11 12:23 PM

Dataset (1) NB| (2) SMO (3) lBK (4) ABM1 (5) J48 (6) PART-------------------------------------------------------------------------------------------------------------------------------mgdata (100) 77.62| 73.71 83.76 78.48 80.69 84.31 -------------------------------------------------------------------------------------------------------------------------------

Our mission is to establish the effectiveness of CI theories by solving industry problems!

“I cannot teach anybody anything, I can only make them think”. – Socrates (470–399 B.C.)

1. Vapnik, V. N., (2005). The Nature of Statitical Learning Theory , Springer.2. Imon, A. H. M. R. (1996). Subsample methods in regression residual prediction and diagnostics. PhD

Thesis, University of Birmingham, UK. 3. Imon, A. H. M. R. (2005). Identifying multiple influential observations in linear regression. Journal of

Applied Statistics, 32, 73 – 90.4. Shafiullah, GM., M. T. Oo, A., Ali, S. A., D. Jarvis, and Wolfs, P., "Prospects of Renewable Energy -

A feasibility study in the Australian context", Accepted for the International Journal of Renewable Energy, ELSEVIER, 2011.

5. Khorshed, M. T., Ali, S., and Wasimi, S., "Monitoring Insiders Activities in Cloud Computing Using Rule Based Learning", Accepted for IEEE TrustCom-11, Nov. 16-18, 2011, Changsha, China.

6. Shafiullah, GM., Ali, S. Thompson, A. and Wolfs, P. "Forecasting Vertical Acceleration of Railway Wagons using Regression Algorithms" IEEE Transactions on Intelligent Transportation Systems, vol. 11, No. 2, June 2010, pp. 290-299.

7. Ali, S. and Pun, D., "Electrofused Magnesium Oxide Classification Using Digital Image Processing and Machine Learning Techniques", Proceeding of The IEEE International Conference on Industrial Technology (ICIT 2009), 10-13 February 2009, Australia.

8. Khorshed, M. T., Ali, S., and Wasimi, S., “A survey on gaps, threats remediation challenges and some thoughts for proactive attack detection in cloud computing “, Submitted to Future Generation Computer System, Elsevier, 2011. (Under Review).

9. Hossain, M. R., M. T. Oo, A., Ali, S.,  " Computational Intelligence: The Effectiveness in Smart Grid ", Submitted to IEEE Transaction on Smart Grid, 2011,

Now your time!

Please ask me your Questions- “?”.