Accounting

Part of Certified Public Accountants (CPAs) work is to assess risk during auditing process. This process of risk assessment involves several procedures each of which analyses data from different sources and which are varying in their degree of reliability. All these analyses are aimed at a single thing risk assessment. Risk assessment is an important tool and is used to assess the probability of fraud occurrence. To simplify this whole process, Certified Public Accountants (CPAs) can use the software known as Neutral Network (NN) that does all this (Green, Reinstein  Harkness, 2001).         

    The purpose of this article is to help practicing accountants use the Neutral Networks better. This will occur only if the CPAs totally understand the Neutral networks. This article aims at educating practicing CPAs on how the Neutral Networks are configured and on how they are programmed to perform various auditing tasks. The article also compares the neutral networks with other recent advances in technological methods of detecting fraud. So as to achieve its objectives, the article sought to answer a number of questions with regard to neutral networks.

    The research questions sought to be answered by this article includes how do neutral networks function Where have they been applied How are neutral networks created What are their advantages and disadvantages After research, the authors of the article were able to come up with answers to these questions.

    To help the CPAs better understand the principle of neutral networks, the article explains that Neutral Networks are systems that ape human way of thinking by applying complex algorithms so as to look at a lot of information simultaneously and make a judgment (Green et al., 2001). In other words they use artificial intelligence. The NN software is installed in a computer and predicts results based on complete or incomplete data. They ape human brain in that they can be trained to identify and understand relationships in different types of complex information which they are able to recall and apply in future.

    Neutral networks have been used to assess the risk of fraud. In addition, they have been used to approve credit, as a stock selection tool as well as to predict bankruptcy (Green et al., 2001). As a tool for approving credit, the Neutral Network system is fed with data and its programmed to detect patterns of credit worthiness or credit unworthiness. When customers apply for credit cards, their financial information is fed to the system and their credit status is determined.  The same procedure is used to predict or determine performance of stocks in the stock market. The NN system is fed with information of low and high performers in the stock market and its trained to predict performance of stocks. Alternatively, the system is fed with data of a particular stocks performance in stock market over a given period of time thereby training it to predict future performance of the stocks (Green et al., 2001). In assessing bankruptcy, the NN system is usually fed with data from collapsed institutions and data from similar institutions that are doing well to train it to discriminate between the two types of scenarios.

    With advanced technology its possible for users to create their own Neutral Networks. Advanced neutral networks software are available that give users the freedom to create prediction and classification tools as they wish. There are also companies that develop and sell specific applications for neutral networks. Its important for a user to know that NN systems differ in design. This determines how usable or suitable a system is for a particular user (Green et al., 2001). Use of the wrong NN system can lead to unreliable and badly interpreted results.

         When well trained, a neutral network is beneficial as its capable of producing consistently good predictions and classifications when fed with real data. Its shortcoming is that it cannot be quantified thus its hard to support the level of precision. Other shortcomings includes the fact that its judgment is not final and so the CPAs still do have to apply their professional judgment, and that to develop the software one needs to have a lot of cases for use in programming the system. Accessing these cases is not always possible.

           To be able to better perform their work, CPAs should consider developing Neutral network systems. These will make their work easier with respect to assessing the risks of fraud and errors on financial statements as well as assessing the going concern of a client. This will also ensure that the limited resources meant for auditing are effectively allocated. However, they should first use the system they develop on trial basis to compare the results the system obtains with those obtained by traditional methods. The point for CPAs to note is that NNs should only be used to supplement the other techniques that are applied in risk assessment in addition to supplementing their professional judgment-they should never be used as a judgment tool in themselves.

        Neutral networks are the long overdue solution to the tedious process of risk assessment. When used well, both the CPAs and their clients can benefit.