Data valuation for medical imaging using Shapley value and application to a large-scale chest X-ray dataset

The reliability of machine learning models can be compromised when trained on low quality data. Many large-scale medical imaging datasets contain low quality labels extracted from sources such as medical reports. Moreover, images within a dataset may have heterogeneous quality due to artifacts and biases arising from equipment or measurement errors. Therefore, algorithms that can automatically identify low quality data are highly desired. In this study, we used data Shapley, a data valuation metric, to quantify the value of training data to the performance of a pneumonia detection algorithm in a large chest X-ray dataset. We characterized the effectiveness of data Shapley in identifying low quality versus valuable data for pneumonia detection. We found that removing training data with high Shapley values decreased the pneumonia detection performance, whereas removing data with low Shapley values improved the model performance. Furthermore, there were more mislabeled examples in low Shapley value data and more true pneumonia cases in high Shapley value data. Our results suggest that low Shapley value indicates mislabeled or poor quality images, whereas high Shapley value indicates data that are valuable for pneumonia detection. Our method can serve as a framework for using data Shapley to denoise large-scale medical imaging datasets.

Measuring the Value of Data and Their Treatment in Macroeconomic Statistics

The paper studies the role of data as an economic asset in the digital economy. The research is focused on the development of an approach to comprehensive data valuation and their adequate treatment in macroeconomic statistics. The first part of the paper reviews the major publications on the so-called Solow productivity paradox: the impact of digital technologies on the productivity growth slowdown. Considering points of view of various researchers, the author takes an opinion that the existing statistical methodology does not permit comprehensive measuring of the digital economy contribution to the productivity dynamics. At the same time, the author does not support the proposal to include the value of data generated by unpaid household activities in macroeconomic accounting and expand the scope of key macroeconomic indicators such as GDP. In the second and the third parts the methods of data valuation used by companies as assets in production, as well as major discussed proposals on methods for measuring the value of data in macroeconomic statistics, are considered. These two aspects of data valuation are closely related, both informationally and methodologically. The author concludes that an increase in the need for the valuation of data at the micro level will inevitably lead to corresponding changes in the methodology of macroeconomic statistics. The last part of the paper explores more elaborately the issues of data valuation as a non-produced asset. The need for such an approach is caused by the existing gap between the marketed assessment of the contribution of data to production and the existing possibilities for accounting for them at the costs of their production. In the author’s opinion, this is a promising direction, allowing to overcome the indicated gap. In support of this, the article provides examples of experimental calculations based on IFRS reports of four Russian companies involved in the production of digital services. Experimental valuation of non-produced assets using the net present value method shows that the value of the non-produced assets involved in the production of data-driven companies differs from the values recorded in their financial statements. This, in particular, occurs due to the underestimation or overestimation of the value of the data used in production, which, according to the author, constitutes the bulk of the unidentified unproduced assets of digital companies. The author concludes that the development of methods for accounting for the value of data as a non-produced asset used in the production of digital products is one of the priority tasks of developing the methodology of the system of national accounts.

Rancang Bangun Penilaian Barang Bergerak Pada Kantor Pt.Mandiri Finance Menggunakan Visual Studio 2010

INTISARIPenilaian adalah proses kegiatan yang dilakukan oleh tim penilai untuk memberikan pendapat berdasarkan data / fakta yang objektif dan relevan dengan menggunakan metode / teknik tertentu pada objek tertentu pada tanggal penilaian. Properti bergerak adalah barang yang sesuai dengan sifat dan penggunaannya dapat dipindahkan. Dalam memproses data penilaian properti bergerak adalah komputerisasi, tetapi hanya menggunakan Microsoft office excel, dan ada kemungkinan besar kesalahan saat memasukkan data atau perhitungan.Dari masalah-masalah ini memunculkan ide membuat aplikasi di mana pemrosesan data, perhitungan, laporan dan pembukuan dapat dilakukan. Bahasa pemrograman yang digunakan adalah Visual Studio 2010. Dimulai dengan memasukkan data tim penilai, data dari objek diikuti oleh data perbandingan sehingga dapat menghasilkan nilai dan melaporkan untuk item tersebut.Dengan aplikasi ini, diharapkan dapat membantu karyawan ketika memasukkan dan menghitung data penilaian barang bergerak, membuat hasil laporan penilaian dan membuatnya lebih mudah untuk mencari data saat dibutuhkan dalam pencarian, memasukkan data dan pelaporan agar lebih efektif dengan pencarian yang lebih baik untuk input data dan pelaporan.Kata kunci – Aplikasi,Barang bergerak ,visual studio 2010 ABSTRACTAssessment Is a process of activities carried out by the assessment team to provide an opinion based on data / facts that are objective and relevant using certain methods / techniques on certain objects at the date of the assessment. Movable property is goods which according to their nature and use can be moved. In processing data valuation of movable property is computerized, but only uses Microsoft office excel, and there is a big possibility of an error when inputting data or calculations.From these problems led to the idea of making an application in which data processing, calculations, reports and bookkeeping can be carried out. The programming language used is Visual Studio 2010. It starts with entering the assessment team data, data from objects followed by comparison data so that it can produce a value and report for the item.With this application, it is expected to be able to assist employees when inputting and calculating moving goods valuation data, making the results of valuation reports and making it easier to search for data when needed in searching, input data and reporting to be more effective with better search for input data and reporting.Keyword -- Application, Moving Property Valuation,visual studio 2010