APPLICATIONS OF MACHINE LEARNING

It may be very clear that the implementation mechanisms associated with machine learning may be responsible towards changes and reengineering the existing specific techniques. Accordingly, the applications towards machine learning may perform better towards specific tasks.

In many times, it may allow the system to learn directly from examples and experience in the form of data, as in this case, being associated with set of tasks along with large amount of data; tasks may be performed by detecting the patterns. In this manner, it may be a much better way to achieve the desired output. In a broad sense, it supports intelligent systems towards particular functions. It also allows the systems to perform specific tasks intelligently, by learning from examples and ignoring the preprogrammed mles.

In general, there are relevant mechanisms associated with machine learning as per following:

  • 1. In supervised machine learning, the trained data may be labeled. Accordingly, each data may be categorized to more groups in different data points. The system may learn about structuring the trained data along with prediction mechanisms.
  • 2. Unsupervised learning may be associated with learning without labels. Sometimes it may aim to detect the characteristics towards making the data points more or less similar to each other.
  • 3. Reinforcement learning, in general, may tend towards unsupervised and supervised learning. In a typical reinforcement learning scheme, the specific entity may interact with its environment and may try to optimize. The primary intention in this case may be focused towards learning strategies.

As an example, machine learning techniques may be associated towards making decision towards diagnosis of symptoms and finding medicines. Preliminary it may allow the system to process written or verbal information towards extraction of information and implement the same towards treatment.

The application of machine learning with specific and large volume of training data may play vital role towards optimizing logistics and associated processes. It may be achieved redirecting towards storage facilities with retrieval mechanisms.

INFORMATION RETRIEVAL (IR)

In very clear term, the system associated with information storage and retrieval may make large volumes of text accessible to the users along with the approaches and goal of the system. The outlines related to the requirements may lead towards relevancy and occurrence of databases along with user’s queries. Accordingly, the response to the queiy may be constructed using the indices and the operations provided by the system associated with conceptualization. It is really essential to model the complexity of the queries as per requirements. But, the users may not know about the indexing terms may be used to describe the databases.

4.4.1 TEXT SYSTEM MODELS ALONG WITH REPRESENTATION

While extracting information from the relevant databases by applying the specific retrieval mechanism, it may be essential to categorize the documents and generate relevancy with accuracy. In this situation, the natural language may have a specific myriad properties used to build the mechanisms towards the system. In fact, the speed of processing is an important factor in large- scale IR, so in reality, only a small set of easily extracted features may be used by any system.

4.4.2 INDEXING WITH SEMANTIC ANALYSIS

In many cases, there may be many ways to be associated with similar objects and it may not be necessary to have literals or words in common. Also, many words may mean quite variant things in different contexts associated with the problems. In fact, the representation may map the queries and documents into an array of significant factor weights leads to specific dimensionality of the new concept of array. The primary concern of the approach may be linked with semantic qualities. Accordingly, the query may be processed by representing its terms as a two-dimensional array in the space and then ranking documents by their proximity to the query tenns. The same may be applied to compare the document clustering with queiy clustering.

4.4.3 SYSTEM ASSOCIATED WITH SEQUENTIAL APPROACH

Towards compressing the required text, many tunes the information may be grouped into strings of arbitrary length. In that scenario, to identify an index, each string may be used to be placed in a compressed version of data associated with the database. Again, to decode the data associated with the database, the hierarchy of the database may be considered and indices may be replaced with appropriate piece of text. In many cases, some specific techniques may be used towards grouping of characters linked to the basis of dividing the text into similar strings.

Now considering the logistic analysis which is nothing but an appropriate analysis towards dependent binary variables. Sometimes it may also be termed as predictive analysis. It may be implemented towards describing data and to explain the relationship between one dependent binary variable and one or more nominal, ordinal, interval or ratio-level independent variables.

STEPS TO MEASURE THE PERFORMANCE OF DATA ACQUISITION AND RETRIEVAL MECHANISM

  • Step 1: Probabilistic measures may be predicted with binary values along with numerical and categorical.
  • Step 2: Values linked to associated and acceptable range also may be considered ranging from 0 to 1.
  • Step 3: Since the experiment may be considered with two possible values, the residuals may not be normally distributed about the predicted line.
  • Step 4: Logistic analysis may be applied using maximum likelihood estimation technique to obtain the coefficients towards relating predictors.
  • Step 5: Evaluation of statistical significance of eveiy coefficient associated with the systems.
  • Step 6: Adoption of testing mechanisms to assess the significance of prediction of each predictor.
  • Step 7: Perform comparison of data along with retrieval mechanisms associated with the system.

APPLICATION OF METAHEURISTIC APPROACH TO THE SYSTEM

As being observed, the metaheuristic algorithms may be used for obtaining contemporary global optimization algorithms, computational intelligence (Cl), and soft computing. These algorithms are in general nature-inspired with multiple interacting agents. The subset of metaheuristics is often termed as swarm intelligence algorithms, linked to swarm intelligence characteristics of biological agents. One of such approach may be termed as firefly algorithm. It is based on the flashing patterns and behavior of fireflies. The attractiveness maybe somehow proportional to the brightness. Accordingly, for any two flashing fireflies, the less bright one will move towards the brighter one. If there is no brighter one than a particular firefly, it will move randomly. Also, the brightness of a firefly may be obtained by the landscape of the objective function (Table 4.1).

TABLE 4.1 Data Servers with Associated Time of Acquisition

SI.

No.

Number of Data Servers

Percentage of Acquiring Data (o/o)

Acquisition Time Associated with Databases (ms)

1.

19

14

0.11

2.

24

19

0.14

3.

30

34

0.37

4.

40

69

0.77

5.

50

83

0.88

Considering Table 4.1, it has been observed that, the percentage of acquiring data may be directly proportional to the size of the server space. Also, the acquisition time associated with the database always depends on the allocation of data servers (Table 4.2).

TABLE 4.2 Data Centers with Associated Time of Acquisition

SI.

No.

Number of Data Centers

Percentage of Acquiring Data (o/o)

Acquisition Time Associated with Data Centers (ms)

1.

19

11

0.09

2.

30

16

0.24

3.

40

47

0.67

4.

50

69

0.88

5.

60

73

0.96

As reflected in Table 4.2, the acquisition time associated with the data centers depends on the allocation of data centers.

Probability of acquisition of data through database

FIGURE 4.1 Probability of acquisition of data through database.

Probabilistic measures of data servers linked with data centers

FIGURE 4.2 Probabilistic measures of data servers linked with data centers.

COMPLEXITY ASSOCIATED WITH THE SYSTEM

It is understood that the metaheuristic algorithms are simple and easy to implement. The main computational cost may be in the evaluations of objective functions. It has attracted much attention and been observed about the computation time for the digital image. It may also produce consistent and better performance in terms of time and optimality. Classifications and clustering are also another important areas of applications, and it has been observed that it may be efficiently used for clustering. Anyway, it may be associated with two distinct merits, i.e., automatic subdivision and ability of dealing with multimodality. The population may be automatically subdivided into subgroups, and each group may swarm around each mode or local optimum. Accordingly, the best global solution may be obtained. After that, the subdivision may permit the fireflies to be able to find all optima simultaneously if the population size may be sufficiently higher than the number of modes.

The virtualized data associated with the servers, in general, may be linked with computing infrastructure, data, and application services. The primary intention in such a case may be to build computing clusters and scale-out the computing capacity. Considering the large set of data in servers, the cost of processing along with computational capacity may be essential. In this regard, while processing the large dataset linked with subsequent models, the changes may be somehow required in the program states along with modularity. In general, being associated with machine learning, the generalized tasks may be synchronized and the workloads may be properly assigned. But somehow, it may not be so effective towards recursive processes.

IMPACT OF MACHINE LEARNING IN VIRTUAL DATABASE

In general, there are basic prerequisites towards virtual databases, i.e., low-cost resources and processing power. Initially, it may be associated with processing power with minimal cost and then the ability to process large size of data. In this regard, the impact of machine learning may be linked with cognitive computing. The storage amount of data in the virtual environment sometimes may be associated with machine learning process. Accordingly, it may provide applications in the virtual environment with sensor capabilities and also the applications may be able to perform the linked functions with suitable decisions. The integration of machine learning may enhance the requirement of virtual space. Also along with suitability of virtual servers, the enhancement of technology may be beneficial towards the relevant fields.

DISCUSSION AND FUTURE DIRECTION

Sometimes, the analysis to predict the requirement, as well as application for both tactical and strategic purposes, may be restricted because of prohibitive resource requirements. But the machine learning associated with the virtual environment may not be constraint towards all large sets of heterogeneous data or data enterprise. Also, the artificial intelligence (AI) systems may be performing better on virtual servers, as linked with low cost of operations, scalability, and huge processing power to analyze the huge amount of data. Accordingly, it may act as a source for the machine LAs. The algorithms associated with machine learning may gather information and perform better with suitability. AI research in a practical situation may result better in a virtual environment.

CONCLUSION

It has been observed that the machine learning technologies along with computation in virtual environment are emerging in the present days. In such a scenario, it may take some time to be folly functional and to be used in crucial sectors like healthcare, business, and banking. It is clear that machine learning may make the data easier to handle over the cloud. Along with repetitive AI research in cloud computing, it may become more and more intelligent. It may be stated that the supervised machine learning models will perform well with a particular dataset, and may perform satisfactorily with totally different datasets generated with different simulation or experimental conditions and environments.

KEYWORDS

  • analytical learning
  • clustering
  • machine learning
  • query terms
  • supervised learning
  • virtual database

REFERENCES

  • 1. Dr. Zakir. L., & Abdulbasit. S., (2013). Web Crawling and Data Mining with Apache Nutch. Packt Publishing.
  • 2. Sanjay, G., Howard. G„ & Shun-Так, L., (2003). “The Google File System" (pp. 1-15). SOSP’03.
  • 3. Stephan, E., Sebastian, S., Kostas, T„ Daniel. W., & Volker. M., (2013). Iterative parallel processing with stratosphere an inside look Proceedings of the 2013 ACM SIGMOD International Conference on Management of Data, SIGMOD 13 (pp 1053-1056).
  • 4. Matei, Z., et al., (2012). Resilient distributed datasets: A fault-tolerant abstraction for in-memory cluster computing. Proceedings of the Ph USENIX Symposium on Networked Systems Design and Implementation.
  • 5. Xin, L., et ah, (2015). Computational Trust Models and Machine Learning. CRC Press.
  • 6. James, S., (2004). The Wisdom of Crowds, Why the Many are Smarter than the Few and How Collective Wisdom Shapes Business, Economies, Societies and Nations (pp. 66-83). Anchor Books.
  • 7. Paul, C. Z., etal., (2013). Harness the Power of Big Data, The IBM Big Data Piatfonn. McGraw-Hill.
  • 8. David. L., Ryan, K.. Gary, K... & Alessandro, V., (2014). The Parable of Google Flu: Traps in Big Data Analysis, Science (VЫ. 343, No. 6176, pp. 1203-1205).
  • 9. Bhat, A. H., Patra, S., & Jena, D., (2013). ‘'Machine learning approach for intrusion detection on cloud virtual machines.” International Journal of Application or Innovation in Engineering and Management (IJAIEM), 2(6), 56-66. http://www.ijaiem. org/Volume2Issue6IJAIEM-2013-06-09-029.pdf (accessed on 25 February 2020).
  • 10. Chase, D. A., Thakar, P., Validat. A., & Doyle, R., (2001). “Managing energy and server resources in hosting centers.” lS^ ACM Symposium on Operating Systems Principles (SIGOPS), 35(5), 103-116. http://dl.acm.org/citation.cfm?id=502045 (accessed on 25 February 2020).
  • 11. Tierney, L., Rossini. A. J., & Na, L., (2009). Snow: A parallel computing framework for the R system. Int. J. Parallel Prog., 37, 78-90. doi: 10.1007/sl0766-008-0077-2.
  • 12. Hido, S., (2012). Jubatus: Distributed Online Machine Learning Framework for Big Data, XLDB Asia. Beijing. http://www.slideshare.net/JubatusOfficial/distributed- onhne-machine-learning-frameworkfor-big-data (accessed on 25 February 2020).
  • 13. Ghoting, A., Kambadur, P., Pednault, E., & Kannan. R. (2011). NIMBLE: A Toolkit for the Implementation of Parallel Data Mining and Machine Learning Algorithms on Map Reduce. KDD 11.
  • 14. Low, Y., Gonzalez, J.. Kyrola, A., Bickson, D., Guestrin, C., & Hellerstein, J. M., (2012). Distributed GraphLab: A framework for machine learning and data mining in the cloud. Proceedings of the VLDB Endowment (Vol. 5, No. 8), Istanbul, Turkey.
  • 15. Pednault. E., Yom-Tov, E., & Ghoting, A., (2012). IBM parallel machine learning toolbox, hi: Bekkerman, R.. Bilenko, M., & Langford, J., (eds.), Scaling up Machine Learning. Cambridge University Press.
  • 16. Eckerson, W„ (2012). New Technologies for Big Data, http://www.b-eyenetwork.com/ blogs/eckerson/archives/2012/ll/newtecknologie.php (accessed on 1 March 2020).
  • 17. Shiravi, A., Shiravi, H., Tavallaee, M., Ghorbani, A., (2012). "Toward developing a systematic approach to generate benchmark datasets for intrusion detection.” Computers and Security, 31(3), 357-374.
  • 18. Sommer. R., & Paxson. V., (2010). “Outside the closed world: On using machine learning for network intrusion detection.” In: IEEE Symposium on Security and Privacy (pp. 305-316).
  • 19. Bhamare. D., Jain. R.. Samaka, M, Vaszkun, G., & Erbad. A., (2015). “Multi-cloud distribution of virtual functions and dynamic service deployment: Open ADN perspective.” In: Cloud Engineering (IC2E) (pp. 299-304). IEEE.
  • 20. Mckugk, J., (2000). “Testing intrusion detection systems: A critique of the 1998 and 1999 DARPA intrusion detection system evaluations as performed by Lincoln Laboratory.” ACM Trans Inf System Security.
 
Source
< Prev   CONTENTS   Source   Next >