3
Artificial Neural Network Applications in Analysis of Forensic Science
1Department of Biotechnology, Sri Padmavati Mahila Visva Vidyalayam (Women’s) University, Tirupati, Andhra Pradesh, India
2Department of Oral Pathology, Saveetha Dental College, Saveetha Institute of Medical and Technical Sciences, Saveetha University, Velappanchavadi, Chennai, Tamil Nadu, India
*Corresponding author: [email protected]
Abstract
Constant growth in crime rates instigates computational resources for examination at a robust rate. Whatever data being examined with the help of forensic tools needs to be stored in the digital memory. Hence artificial intelligence is the upcoming machine learning technology which is comprehensive for human minds and provides capacity of digital storage media which can be accessed when in need. The purpose of our current research is to have broader understanding about the applicability of Artificial Intelligence (AI) along with computational logic tools analysis. The present artificial neural network helps in detection of criminals through comparison of faces by employing deep learning which offers neural networks. Thus our paper focus on the computational forensic approaches built with AI applications to detect and predict possible future crimes. Several in-built algorithms control and create a model image in a camera which can be utilized in forensic casework to solve cases robustly.
Keywords: Artificial intelligence, computational logic tools, artefacts sensor, algorithms, digital memory, forensic tools, deep learning
3.1 Introduction
In the present era, artificial intelligence is materializing as the utmost vital science in all facets of life. Similarly, forensic science is also being acknowledged due to the advancement in machine learning, deep learning and natural language processing of science. Today, our country is mostly digitalized because of the impact of AI technology [1]. Revolution in Information Technology and progression in telecommunications with the help of the Internet especially has been found to benefit forensic examination along with preservation of information related to crime and further analysis of those evidences when necessary [2, 3].
The AI-influenced artificial neural networks (ANN) are employed in forensic science analysis for predicting crimes. In forensic science for analysis of an offence the application of ANNs provides precision, practiced faster even with small volume of sample [4, 5]. The implementation of ANNs is the most widely highlighted technology in the forensic research domain [6]. Figure 3.1 depicts the progression of Artificial Neural Networks in detection of patterns for personal identification through deep learning processing. The biological neural network of signalling pattern begins from dendrites, where it receives its input messages and starts its processing/predicting the hidden patterns inside the brain and finally outputs through the synapses. However, the same networking system is employed by artificial intelligence which has been regarded as “Intelligence agents” which is very identical to and even more knowledgeable than the human brain [7].
Figure 3.1 Artificial neural network works as patterns in predicting information.
Computerized Forensics science is a subdivision of criminology which deals with the recognition, execution, maintenance, scrutiny and organization of the evidence matter in digitalized systems, which is generally regarded as an automatic data processing system. Precisely, our major focus in this review article is the employment of AI-built digital devices for evidence analysis during the crime scene which has been built with Answer Set Programming (ASP) which decodes through computational logic paradigm [8–11]. Hence, harmonious exploration of AI devices in forensic analysis helped to solve corruption at a robust rate.
3.2 Digital Forensic Analysis Knowledge
It is evident that medical inspection with digital devices can help in the speedy gathering of proofs to solve cases. For instance, in reports of cyber-café where we find similar IP address in all the systems utilized. However, the same applies in the case of fraud involving companies which also have many digital devices and users. Computerized forensic analysis examines the wrongdoing by software tools such as Encase, forensic tool kits (FTK) which is a guidance software for reasoning and reporting corruption [12, 13]. Hence, in the field of criminology/forensic science the application of deep learning process have provided more cognitive information to predict even the future happenings of corruption at a very rapid rate. Therefore, the artificial intelligence technology usage in cyber forensics led to framing of deep learning cognitive computing (DLCF) for solving the problems related to crime. (Shown in Figure 3.2).
3.3 Answer Set Programming in Digital Investigations
Digital Forensics (DF) is that domain of science specifically dealing with corruption and it provides accurate information for identification, preservation, extraction and finally decision which highlights the relevant documents by answering all the required queries laid during the investigation phases. The computer is programmed with polynomial hierarchy with answer set programming paradigm via interference engine or ASP problem solver [14, 15]. This ASP is built with respect to vital languages and has the ability to easily read, edit the text file with ASP rules along with analysis of huge data at a faster rate. However, at present the enhancing artificial intelligence technology with natural language processing plus answer set programming can synergistically uplift this machinery application in forensic science.
Figure 3.2 Deep learning enabled cyber forensic investigation analysis.
In order to solve the crime the matrix has to be constructed which is the probable path that exists for detection of commitment of the offense and is further predicted with the help of ASP software. Inside the matrix the neglected cells are presumed to have a number 0 and here we have employed ‘clingo’ solver.
Matrix (1, 1, 18). Matrix (1, 5, 26). Matrix (2, 1, 19). Matrix (2, 4, 27). Matrix (3, 2, 14). Matrix (3, 5, 23). Matrix (3, 6, 31). Matrix (4, 1, 1). Matrix (4, 4, 8). Matrix (4, 5, 33). Matrix (5, 3, 5). Matrix (6, 3, 10). Matrix (6, 5, 36). Matrix (6, 6, 35).
The conundrum of Hidato is derived from the Hebrew word “Hida,” which means puzzle/mystery, logical brainteaser (also regarded as “Hidoku”) designed by the arithmetician Dr. Gyora Benedek from Israel. However, the major purpose of Hidato was to satiate the pattern with numerics horizontally, vertically or diagonal ideal line. To predict the crime scene consideration of the matrix in Figure 3.3 is crucial. For an assumption if the corruption has taken place at the cell space spotted with 0 located which is amongst 14, 8 and 5 at that point we need to take the hiatus with lower bound analogous when the dubious was at position 1 and higher compelled corresponding to when the dubious was at locality 36. Therefore, all stratagems have been undoubtedly swapped off when lengthy zero’s series happens [16].
3.4 Data Science Processing with Artificial Intelligence Models
Nevertheless, data/material science advancement with accessibility of large datasets conglomerate with the expansion in algorithms, plus upsurging progression in computing programming, kindled curiosity for readers to gain knowledge, especially on forensic analysis with the help of AI-constructed machines built for high-dimensional output of data. Moreover, the machine learning was ascertained to have phenomenal aptitude in various fields like image processing, video games, automatic car driving, voice recognition, IP address detection, Spam detection, fraudulent web searches, etc. [17–23]. (Shown in Figure 3.4).
3.5 Pattern Recognition Techniques
The latest technique for analysis of crime to a robust extent is the key quality in forensic science, since this data science progression, which is a subgroup of criminology detection, is practiced for uncovering diverse patterns/image forms of trends from huge data. The pattern identification tendency is instituted based upon concrete evidence and probabilistic thinking. Hence, AI, ANN has been regarded as the most effectual technique in the recognition of these trends from convoluted data. Some examples are in this article for highlighting the AI brain in pattern detection models (model presented in Figure 3.5). The artificial neural network pattern programmed with machine learning, deep learning algorithm attempts to spot sundry parts of a portrait or an individual [24, 25]. Moreover, specific modes for pattern recognition are there, like recognition of spams patterns in email differs from configuration of sound, and similarly fingerprint patterns are also diverse, but all big data obtained helps in firm identification of patterns with a high degree of performance output through artificial neural networking programmes constructed with artificial intelligence technology. Hence, the employment of Artificial Intelligence/ANN can often mitigate the levels of false positive or false negative outcomes [26].
Figure 3.3 Hidato puzzle (Hidoku) matrix list (Kjellerstrand, 2015).
Figure 3.4 Role of artificial intelligence in data science.
Figure 3.5 Model for pattern recognition in forensic analysis.
3.6 ANN Applications
The artificial neural network has been utilized widely in all disciplines like educational purpose, in economics, and in forensic science for detection of criminals, predicting the scene of crime, and today we are facing the coronavirus pandemic where ANN plays a key role in predicting drugs for diagnosis. Our paper has focussed on neural networks performance and its application to the global threat. This ANN technology works with machine learning plus deep learning programs for solving problems. Nevertheless, several researchers are tremendous identifiers of data patterns plus predicting shares in business/forecasting etc. [27–35]. For instance, highlights about ANN in disease prediction are depicted in Table 3.1.
3.7 Knowledge on Stages of Digital Forensic Analysis
The knowledge of forensic science is simplified with the introduction of digital forensics which helps in solving problems during complex investigation procedures. The main phases of forensic science are depicted in Figure 3.6 which includes 1) the Recognition phase: The goal of this phase is to identify which are reliable proofs to be collected and stored; 2) Acquisition phase: The second phase is the chief stage for gathering all relevant evidence in connection with crime; 3) Perpetuation: All the technical actions are considered and managed during trial phases; 4) Investigation: Based on hypothesis in combination with scientific methodology the aim is to confirm/disprove the crime; 5) Evidence: Nevertheless, the final stage is critically intended to record the actions and outcomes through formal reports.
Table 3.1 Application of artificial neural networking in predicting diseases.
| S. no | Entitlement | Technique | Prediction outcome |
| 1 | Derive data sets from appropriate site and merged with neural networks based on hierarchy for recognition of cardiovascular diseases [36]. | Fuzzy neural network method/algorithm employed | Analysis of variance is based upon the outcome and acknowledged by their characteristic features |
| 2 | Identification of Diabetes Mellitus by ANN’s [37]. | Algorithm used is Back Propagation | The best output performance is 82% |
| 3 | Diabetes Mellitus is predicted with artificial neural networks [38]. | Regression plots method | Accuracy based on Bayesian regulation and exhibiting 88.9% |
| 4 | Neonatal disease diagnosis utilization of ANNs [39]. | Multilayer Back propagation algorithm | Predicting accuracy achieved was 75% |
| 5 | Incidence of Salmonellosis forecasted with help of ANNs [40]. | Algorithm used was back propagation | The empirical result was based on Theil’s U with a value of 0.209 |
Figure 3.6 Phases of digital investigation provides knowledge in forensic science analysis.
Thus the knowledge of stages in forensic science provides identification of any corruption from small fragments into a transformed proof to be presented for trial [41–43].
3.8 Deep Learning and Modelling
The non-linear projection formation is based upon functions similar to the nervous system of the brain, i.e., neurons in transmission of impulses. The Artificial Neural Networks are regarded as the most effective device for modelling, particularly when the data connexion is mysterious or unfamiliar. Hence, ANNs can detect and study interconnected patterns between input data sets and resultant target values. The first and foremost necessity is to train the ANNs to envisage the aftermath of any latest entered data. Artificial Neural Networks employs deep erudition method similar to that of the humanoid brain and can solve glitches even from non-linear multifaceted information. Moreover, ANNs look very natural and identical to that of human neurons in the brain. Although the interconnected neurons help to read complicated delinquent with much ease, the computational structure of ANNs with densely connected processing unit helps to predict/forecast the occurrences. Figure 3.7 shows the general modelling pattern of ANN with respect to human nervous system, i.e., neurons.
Figure 3.7 Pattern recognition in deep learning algorithm.
The neural network has to be fed with messages and once the messages enter, the brain automatically starts connecting with nodes which are regarded as hidden layers inside the brain. Since it is a computational biology, in order to predict the outcome it requires a few algorithms such as feed forward neural network, back propagation and regression plots. Therefore, the ANNs modelling with deep learning program have wider application in many disciplines like pattern recognition, power systems, 5G robots in control of pandemic outbreaks, forecasting, manufacturing, social sciences and psychological sciences signal processing [44–51].
3.9 Conclusion
The basic challenge of this paper is to provide readers with a conceptual idea about the upsurging artificial astuteness and computerized reasoning in the digital forensics investigation to find solutions to undefined problems. At present, AI technology has created a revolution globally with automated specific modules to perform tasks with much ease. This review is based on the AI perspective which aims to construct software tools with complex connexions from diverse fields such as diagnosis, forecasting, prediction, temporal learning with logical reasoning and finally conceptual analysis. Therefore, the intelligence agents built with automated tools help in digital investigation and solve cases robustly in comparison to exhaustive searches conducted by human observation. Thus, artificial neural networking with the trend of pattern recognition became a breakthrough and is helpful in the solving of crime scenarios.
References
1. Turner, P., Unification of digital evidence from disparate sources (Digital Evidence Bags). Digital Investigation 2(3): 223-228, 2005.
2. Hudson, P., Why the tech revolution is the industrial revolution of our time, Available at: https://www.elitedaily.com/news/technology/tech-revolutionindustrial-revolution-time, 2013, Accessed 15 February 2019.
3. Kaufman, M., The internet revolution is the new industrial revolution, Available at, https://www.forbes.com/sites/michakaufman/2012/10/05/theinternet-revolution-is-the-new-industrial-revolution/#22b4783447d5, 2012. (Accessed 15 February 2019).
4. Garcia, H. He, E.A., Learning from imbalanced data, IEEE Trans. Knowl. Data Eng. 21 (9) 1263-1284, 2009.
5. Mozaffari, A., Emami, M., Fathi, A., A comprehensive investigation into the performance, robustness, scalability and convergence of chaos-enhanced evolutionary algorithms with boundary constraints, Artif. Intell. Rev. 1-62, 2018.
6. Izeboudjen, N., Larbes, C., Farah, A., A new classification approach for neural networks hardware: from standards chips to embedded systems on chip, Artif. Intell. Rev. 41 (4) 491-534, 2014.
7. Richard P., Lipmann, “An Introduction to Computing with Neural Nets” (IEEE ASSP Magazine, April 1987.
8. Alberti, M., Gavanelli, M., Lamma, E., Mello, P., Torroni, P, The Sciff abductive proof-procedure. In: Bandini, S., Manzoni, S. (eds.) AI*IA 2005: Advances in Artificial Intelligence, 9th Congress of the Italian Association for Artificial Intelligence, Proceedings, volume 3673 of Lecture Notes in Computer Science, pp. 135–147, Springer, 2005.
9. Kakas, A.C., Kowalski, R.A., Toni, F.: Abductive logic programming. J. Log. Comput. 2(6), 719–770, 1992.
10. Kakas, A.C., Toni, F, Computing argumentation in logic programming. J. Log. Comput. 9(4), 515–562, 1999.
11. Niemela, I., Logic programs with stable model semantics as a constraint programming paradigm. Ann. Math. Artif. Intell., Springer 25(3-4), 241–273, 1999.
12. Zhang, Y., Lin, Y., Research on the key technology of secure computer forensics in: 2010 Third International Symposium on Intelligent Information Technology and Security Informatics, pp. 649-652, 2010.
13. Ling, T., June., The study of computer forensics on linux, in: Computational and Information Sciences (ICCIS), 2013 Fifth International Conference on, IEEE, pp. 294-297, 2013.
14. Maratea, M., Pulina, L., Ricca, F. Multi-engine ASP solving with policy adaptation. J. Log. Comput. 25(6), 1285–1306, 2015.
15. ASP. Answer set programming solvers online (incomplete list), http://assat.cs.ust.hk, http://www.cs.utexas.edu/users/tag/ccalc/, http://www.cs.utexas.edu/users/tag/cmodels/, http://www.cs.uky.edu/ai/, http://www.dbai.tuwien.ac.at/proj/dlv, http://www.potassco.org, http://www.tcs.hut.fi/Software/smodels. 2016.
16. Kjellerstrand, H., Hidato., Available at http://www.hakank.org/answer set programming 2015.
17. Marsland, S., Machine Learning, CRC Press, Taylor & Francis Inc., Boca Raton, FL, 2014.
18. Silver, D. et al., Mastering the game of go with deep neural networks and tree search. Nature 529, 484–489, 2016.
19. Bojarski, M. et al., End to end learning for self-driving cars. Preprint at arXiv:1604.07316, 2016.
20. He, K., Zhang, X., Ren, S., & Sun, J., Delving deep into rectifiers: surpassing human level performance on ImageNet classification. In 2015 IEEE International Conference on Computer Vision (ICCV) (Eds Bajcsy, R. & Hager, G.) 1026–1034 (IEEE, Piscataway, NJ, 2015.
21. Liu., S.-S., & Tian, Y.,-T, Facial expression recognition method based on gabor wavelet features and fractional power polynomial kernel PCA. In Advances in Neural Networks - ISNN 2010 (Eds Zhang, L., Lu, B.-L. & Kwok, J.) 144–151 (Springer, Berlin, Heidelberg, 2010.
22. Waibel, A. & Lee, K.-F. (Eds) Readings in Speech Recognition, Morgan Kaufmann, Burlington, MA, 1990.
23. Pazzani, M., & Billsus, D., Learning and revising user profiles: the identification of interesting web sites. Mach. Learn. 27, 313–331, 1997.
24. X. Zhou, W., Gong, W., Fu, F. Du, Application of deep learning in object detection, in: IEEE/ACIS 16th International Conference on Computer and Information Science (ICIS), IEEE, pp. 631-634, 2017.
25. M. Mahmud, M.S. Kaiser, A. Hussain, S. Vassanelli, Applications of deep learning and reinforcement learning to biological data, IEEE Trans. Neural Network. Learn. Syst. 29 (6) 2063-2079, 2018.
26. Bishop, C.M., 1995. Neural Networks for Pattern Recognition, Oxford University Press.
27. S.S. Haykin, Kalman Filtering and Neural Networks, Wiley, New York, pp. 221-269, 2001.
28. Zhang, W. Yu, H. He, N., Advances in Neural Networks, ISNN 2009 6th International Symposium, 2009.
29. Fan, W., Bouguila, N., Ziou, D., Variational learning for finite Dirichlet mixture models and applications, IEEE Trans. Neural Network. Learn. Syst. 23 (5) 762-774, 2012.
30. Saravanan, K., Sasithra, S., Review on classification based on artificial neural networks, Int. J. Ambient Syst. Appl. (IJASA) 2 (4) 11-18, 2014.
31. Martínez-Porchas, M., Villalpando-Canchola, E., Vargas-Albores, F., Significant loss of sensitivity and specificity in the taxonomic classification occurs when short 16S rRNA gene sequences are used, Heliyon 2 (9) e00170, 2016.
32. Abid, F., Hamami, L., A survey of neural network based automated systems for human chromosome classification, Artif. Intell. Rev. 49 (1) (2018) 41-56, 2018.
33. Das, K., Behera, R.N., A survey on machine learning: concept, algorithms and applications, Int. J. Innovat. Res. Comput. Commun. Eng. 5 (2), 1301-1309, 2017.
34. R. Boutaba, M.A., Salahuddin, N., Limam, S., Ayoubi, N., Shahriar, F., EstradaSolano, O.M., Caicedo, A., comprehensive survey on machine learning for networking: evolution, applications and research opportunities, J. Internet Serv. Appl. 9 (1) 16, 2018.
35. Ogwueleka, F.N., Misra, S., Colomo-Palacios,R., Fernandez,L., Neural network and classification approach in identifying customer behavior in the banking sector: a case study of an international bank, Hum. Factors Ergon. Manuf. Serv. Ind. 25 (1) 28-42, 2015.
36. Jun Shi., Ming Chui, D., “Extract Knowledge from Site-sampled Data Sets and Fused Hierarchical Neural Networks for Detecting Cardiovascular Diseases International Conference on Biomedical Engineering and Biotechnology, 2012.
37. Acar, E., Özerdem, M.S., and Akpolat,V., “Diabetes Mellitus Forecast Using Various Types of Artificial Neural Network,” International Advanced Technologies Symposium (IATS’11), 2001.
38. Sapon, M.A., Ismail, K., and Zainudin, S., “Prediction of Diabetes by using Artificial Neural Network”, International Conference on Circuits, System and Simulation IPCSIT vol. 7, 2011.
39. Chowdhury, D.R., Chatterjee, M., & Samanta R.K., “An Artificial Neural Network Model for Neonatal Disease Diagnosis”, International Journal of Artificial Intelligence and Expert Systems (IJAE), Volume (2): Issue (3), 2011.
40. Permanasari, A. E., Awang Rambli, D.R., and Dominic, P.D.D., “Forecasting of Salmonellosis Incidence in Human using Artificial Neural Network (ANN),” presented at 2nd International Conference on Computer and Automation Engineering (ICCAE) Volume 1, Singapore, pp. 136-139, 2010.
41. Kumar, G., Artificial Neural Networks JHA, Indian Agricultural Research Institute, PUSA, New Delhi, 2012.
42. Atiya, A.F., El-Shoura, S.M., Shaheen,S.I., and Sherif, M.S. EI., “A Comparison Between Neural-Network Forecasting Techniques—Case Study: River Flow Forecasting,” IEEE Transactions on Neural Networks, vol. 10, pp. 402-409, 1999.
43. Ludermir, T.B., Yamazaki, A., Zanchettin, C., An optimization methodology for neural network weights and architectures, IEEE Trans. Neural Network. 17 (6) 1452-1459, 2006.
44. Gopalapillai, R., Vidhya, J., Gupta,D., Sudarshan, T.S.B., Classification of robotic data using artificial neural network, in: Intelligent Computational Systems (RAICS), IEEE Recent Advances in (pp. 333-337), IEEE, December 2013.
45. Kaminski, W., Skrzypski, J., Jach-Szakiel, E., Application of artificial neural networks (ANNs) to predict air quality classes in big cities, in: 19th International Conference on Systems Engineering (pp. 135-140), IEEE, August 2008.
46. Wang, S.C., Dong, J.X., Shen, G., ANN-based process control in manufacturing, in: American Control Conference, IEEE, pp. 2531-2532, 1993.
47. Pardo, T. Le, P., Claster, W., Application of artificial neural network in social media data analysis: a case of lodging business in Philadelphia, in: Artificial Neural Network Modelling (pp. 369-376), Springer, Cham, 2016.
48. Barni, M., A. Pelagotti, A., Piva,A., Image processing for the analysis and conservation of paintings: opportunities and challenges, IEEE Signal Process. Mag. 22 (5) 141-144, 2005.
49. Li, S. He, X., Application of a group search optimization based artificial neural network to machine condition monitoring, in: Emerging Technologies and Factory Automation, 2008. ETFA 2008. IEEE International Conference on (pp. 1260-1266), IEEE, September 2008.
50. Barto, A.G., Sutton, R.S., Anderson, C.W., Neuron like adaptive elements that can solve difficult learning control problems, IEEE Trans. Syst. Man Cybern. (5) 834-846, 1983.
51. Bogdan, M., Schroder, M., Rosenstiel, W., Artificial neural net based signal processing for interaction with peripheral nervous system, in: Neural Engineering, Conference Proceedings. First International IEEE EMBS Conference on (pp. 134-137), IEEE, March 2003.