During the financial crisis, the delayed recognition of credit losses on loans and other financial instruments was identified as a weakness in existing incurred loss model of impairment stated by International Account...During the financial crisis, the delayed recognition of credit losses on loans and other financial instruments was identified as a weakness in existing incurred loss model of impairment stated by International Accounting Standards （IAS） 39, because it is believed that this delay might generate pro-cyclical effects. In response to the recommendations of G20, Financial Crisis Advisory Group （FCAG）, and other international bodies, the International Accounting Standards Board （IASB） has undertaken, since 2009, as a part of the project to replace IAS 39, a project （partially shared with Financial Accounting Standards Board （FASB）） aimed at introducing an expected loss model of impairment. Within the scope of this subset project, the IASB has previously issued two exposure documents proposing models to account for expected credit losses： an exposure draft （ED） Financial Instrument： Amortized Cost and Impairment, published in November 2009, and a supplementary document （SD） Financial Instrument： Impairment, published jointly with the FASB in January 2011. However, neither of the two proposals received strong support from interested parties. Recently, the IASB, after the FASB＇s decision to withdraw from the joint project and to develop a separate expected credit loss model based on a single measurement approach consisting in the sole recognition of lifetime expected credit losses, published a third proposal--Ahe so-called expected credit losses model （ED/2013/3 Financial Instruments： Expected Credit Losses）.展开更多
Accuracy of machine learners is affected by quality of the data the learners are induced on. In this paper, quality of the training dataset is improved by removing instances detected as noisy by the Partitioning Filte...Accuracy of machine learners is affected by quality of the data the learners are induced on. In this paper, quality of the training dataset is improved by removing instances detected as noisy by the Partitioning Filter. The fit dataset is first split into subsets, and different base learners are induced on each of these splits. The predictions are combined in such a way that an instance is identified as noisy if it is misclassified by a certain number of base learners. Two versions of the Partitioning Filter are used： Multiple-Partitioning Filter and Iterative-Partitioning Filter. The number of instances removed by the filters is tuned by the voting scheme of the filter and the number of iterations. The primary aim of this study is to compare the predictive performances of the final models built on the filtered and the un-filtered training datasets. A case study of software measurement data of a high assurance software project is performed. It is shown that predictive performances of models built on the filtered fit datasets and evaluated on a noisy test dataset are generally better than those built on the noisy （un-filtered） fit dataset. However, predictive performance based on certain aggressive filters is affected by presence of noise in the evaluation dataset.展开更多
文摘During the financial crisis, the delayed recognition of credit losses on loans and other financial instruments was identified as a weakness in existing incurred loss model of impairment stated by International Accounting Standards （IAS） 39, because it is believed that this delay might generate pro-cyclical effects. In response to the recommendations of G20, Financial Crisis Advisory Group （FCAG）, and other international bodies, the International Accounting Standards Board （IASB） has undertaken, since 2009, as a part of the project to replace IAS 39, a project （partially shared with Financial Accounting Standards Board （FASB）） aimed at introducing an expected loss model of impairment. Within the scope of this subset project, the IASB has previously issued two exposure documents proposing models to account for expected credit losses： an exposure draft （ED） Financial Instrument： Amortized Cost and Impairment, published in November 2009, and a supplementary document （SD） Financial Instrument： Impairment, published jointly with the FASB in January 2011. However, neither of the two proposals received strong support from interested parties. Recently, the IASB, after the FASB＇s decision to withdraw from the joint project and to develop a separate expected credit loss model based on a single measurement approach consisting in the sole recognition of lifetime expected credit losses, published a third proposal--Ahe so-called expected credit losses model （ED/2013/3 Financial Instruments： Expected Credit Losses）.
文摘Accuracy of machine learners is affected by quality of the data the learners are induced on. In this paper, quality of the training dataset is improved by removing instances detected as noisy by the Partitioning Filter. The fit dataset is first split into subsets, and different base learners are induced on each of these splits. The predictions are combined in such a way that an instance is identified as noisy if it is misclassified by a certain number of base learners. Two versions of the Partitioning Filter are used： Multiple-Partitioning Filter and Iterative-Partitioning Filter. The number of instances removed by the filters is tuned by the voting scheme of the filter and the number of iterations. The primary aim of this study is to compare the predictive performances of the final models built on the filtered and the un-filtered training datasets. A case study of software measurement data of a high assurance software project is performed. It is shown that predictive performances of models built on the filtered fit datasets and evaluated on a noisy test dataset are generally better than those built on the noisy （un-filtered） fit dataset. However, predictive performance based on certain aggressive filters is affected by presence of noise in the evaluation dataset.
