This document provides an overview of neural networks and their potential applications in accounting and auditing. It discusses how neural networks work, their history of use since the 1990s, and current applications in areas like continuous auditing, fraud detection, and improving auditor decisions. While neural networks have seen limited adoption in accounting and auditing so far, the document argues they could benefit the field by identifying patterns in large datasets that humans may miss. It recommends auditing professionals implement neural network models with a full-time commitment to help direct their work.

1. Matthew Chan Section 1 ACC 626 Neural Networks in Accounting and Auditing

2. Audio Synchronization Due to technical difficulties with Slideshare, it will be necessary to change slides manually, I apologize for this inconvenience. The slides are as follows:

3. Link to MP3 http://www.archive.org/details/NeuralNetworksSlidecastMp3

4. Agenda Introduction Background How it Works History Current State Cost Effective? Recommendations Conclusion

5. Introduction Neural Networks not widely used in audit Used mainly in science, and has some applications for fraud detection Has been tested and proposed for various uses in accounting and auditing

6. Background Neural networks are a type of artificial intelligence, and are based on the structure of the human brain and are composed of a large number of interconnected processors Pattern recognition is one of the most important aspects of the neural network technology The main advantage of artificial neural networks is that they can learn from their inputs and examples that are inputted into them can see relationships in data that will not be noticeable to human observers

7. How it Works

8. History In 1994, neural networks were a new type of technology In 2003, large companies began to implement neural networks Robert Hecht-Nielsen, a professor at the University of California, San Diego, called neural networks “the most important scientific challenge of our time”

9. Current State of Neural Networks in Auditing Neural networks have been used in various applications across the scientific world, but they have not seen widespread implementation in the areas of accounting and auditing This section will examine the application of neural networks in various areas

10. Continuous Auditing Continuous auditing is defined as a type of auditing which produces audit results simultaneously, or a short period of time after, the occurrence of relevant events Internal auditors have a very prominent role in continuous auditing of a company Training efforts should have sufficient depth

11. Fraud Detection Neural networks have become the method of choice in the realm of fraud detection For example, a study used a neural network system to identify possible areas in financial data that would lead to fraud lawsuits The model will be able to assess this probability of fraud litigation

12. Accounting Buried deep within accounting data are patterns Accounts that traditionally are not viewed as having strong correlations with each other may in fact present substantial relationships

13. Auditor Decisions Many parts of the auditing field have been subject to neural network testing The two duties that stand out in terms of importance are the evaluation of a going concern opinion, and issuing qualified opinions in audit reports

14. Going Concern Being able to forecast earnings would allow for auditors to see if a company is likely to be able to continue as a going concern, and neural networks can establish patterns to see financial viability in the future No matter how many inputs are put into the system, and what the system generates as an output, it is still ultimately the auditor’s decision in the end

15. Qualified Opinions Being able to determine predictive patterns with regards to qualified audit opinions would allow auditors to “plan specific auditing procedures to achieve an acceptable level of audit risk” Large amounts of data could be used to focus auditors’ attention on possible problem areas that may result in a qualified opinion, and allow for a greater degree of testing and analysis with regards to these problem areas

16. Improvements? Their current lack of usage by the auditing profession should be reassessed The large amount of studies being performed over the past ten years is evidence of a growing number of academics and professionals who believe that this technology can provide great benefits to the auditing community

17. Cost Effective? It will become increasingly necessary to keep up by employing neural networks to observe patterns that are beyond the ability of human auditors Employing neural networks would also lead to a greater degree of accuracy in dangerous situations If auditors are not able to have a tool on their side that can generate predictive data about these issues, then they may not be able to keep pace with their own profession

18. Recommendations It is recommended that in the presence of evolving technology, a neural network model should be implemented into usage by auditing professionals Using the system would likely require a full-time commitment The neural network would have to be applied to each audit on an individual basis, so once a firm implements the necessary infrastructure for the network, individual audits can be analyzed

19. Conclusion Their specific application to various issues within the auditing world, such as continuous auditing, fraud detection, auditor decision making (going concern evaluation as well as issuing qualified audit opinions) could change the entire profession It is important for the auditing world to see that neural networks are not a replacement for the expertise and professional judgment of auditors, but simply a means of directing their attention and recognizing patterns in large amounts of data that humans would not see

20. Works Cited Vasarhelyi, Miklos A., and Alexander Kogan. "Part II." Artificial Intelligence in Accounting and Auditing. Vol. 4: towards New Paradigms. Vol. 4. Princeton, NJ: Markus Wiener Pub, 1997. 64. Print. Garrity, Edward J., Joseph B. O'Donnell, and G. Lawrence Sanders. "Continuous Auditing and Data Mining." Idea Group, 2006. Web. 25 May 2011. <2. http://www.irma-international.org/viewtitle/10596/>. Aparaschivei, Florin. "A Conexionist Intelligent System for Accounting." RevistaInformaticaEconomica, 2008. Web. 25 May 2011. <http://revistaie.ase.ro/content/45/11%20-%20Florin_Aparaschivei.pdf>. Bhattacharya, Sukanto, DongmingXu, and Kuldeep Kumar. "An ANN-based Auditor Decision Support System Using Benford's Law." Decision Support Systems 50 (2011): 576-84. Print. Chen, H., S. Huang, and C. Kuo. "Using the Artificial Neural Network to Predict Fraud Litigation: Some Empirical Evidence from Emerging Markets." Expert Systems with Applications 36.2 (2009): 1478-484. Print. Gaganis, C., F. Pasiouras, and M. Doumpos. "Probabilistic Neural Networks for the Identification of Qualified Audit Opinions." Expert Systems with Applications 32.1 (2007): 114-24. Print.

21. Works Cited Pasiouras, F., C. Gaganis, and C. Zopounidis. "Multicriteria Decision Support Methodologies for Auditing Decisions: The Case of Qualified Audit Reports in the UK." European Journal of Operational Research 180.3 (2007): 1317-330. Print. Kirkos, E., C. Spathis, and Y. Manolopoulos. "Data Mining Techniques for the Detection of Fraudulent Financial Statements." Expert Systems with Applications 32.4 (2007): 995-1003. Print. Stefanou, Constantinos J. "The Complexity and the Research Area of AIS." Journal of Enterprise Information Management 19.1 (2006): 9-12. Print. Verschoor, Curtis C. "Continuous Auditing: An Operational Model for Internal Auditors." Internal Auditing 21.2 (2006): 43-44. Print. Warren Jr., J Donald, and L Murphy Smith. "Continuous Auditing: An Effective Tool for Internal Auditors." Internal Auditing 21.2 (2006): 27-35. Print. Etheridge, Harlan L., Ram S. Sriram, and H. Y. Kathy Hsu. "A Comparison of Selected Artificial Neural Networks That Help Auditors Evaluate Client Financial Viability." Decision Sciences 31.2 (2000): 531-50. Print.

22. Works Cited Etheridge, Harlan L., and Ram S. Sriram. "A Comparison of the Relative Costs of Financial Distress Models: Artificial Neural Networks, Logit and Multivariate Discriminant Analysis." International Journal of Intelligent Systems in Accounting, Finance & Management 6.3 (1997): 235-48. Print. Etheridge, Harlan L., and Richard C. Brooks. "Neural Networks: A New Technology." The CPA Journal 64.3 (1994): 36+. Print. Cerullo, Michael J., and M. Virginia Cerullo. "Using Neural Network Software as a Forensic Accounting Tool." ISACA Journal (2006): 1-5. Print. Hoover, J. Nicholas. "ABOUT FACE; CA's Alive with New People, Products, and Practices, but Old Habits-and Impressions-die Hard." INFORMATIONWEEK (2006): 1-8. Print. Koprowski, Gene J. "Technology News: News: Neural-Network Technology Moves into the Mainstream." TechNewsWorld: All Tech - All The Time. Web. 20 June 2011. <http://www.technewsworld.com/story/31280.html?wlc=1308456992>.

23. Works Cited "Probabilistic and General Regression Neural Networks." Probabilistic and General Regression Neural Networks. Web. 20 June 2011. <http://www.dtreg.com/pnn.htm>. "Interference and Old Data." SCHOOL OF COMPUTER SCIENCE, Carnegie Mellon. Web. 23 June 2011. <http://www.cs.cmu.edu/~schneide/tut5/node28.html>. "What Are the Pros and Cons of Neural Networks from a Practical Perspective?" Quora. Web. 25 June 2011. <http://www.quora.com/What-are-the-pros-and-cons-of-neural-networks-from-a-practical-perspective>.