去雜訊相關性方法在反推問題上的可靠性

Abstract

現今，在許多領域中都存在著大量的量測數據，然而造成這些數據的底下機制卻往往很難求得，因此如何從這些含有雜訊的數據反推原來龐大的網絡結構就成為一項重要的議題。傳統擬合的方法在處理高維系統時甚為繁瑣，為了解決這個複雜問題，最近出現了一些新想法。此論文使用這些新發展的理論，在不同動態系統下觀察反推的結果，並針對現有的方法討論其適用性，及擴展理論的可能。最後我們嘗試將此理論應用到生物系統，例如使用在簡單的細胞偵測外界訊號分子濃度變化機制的生物問題上。

Nowadays, massive amounts of measured data are available for analysis in various fields. However, the underlying mechanism yielding these data are often hard to extract. Therefore, it has become an important issue how to inversely deduce the mechanism of these data. Calculations by typical fitting methods will be rather complicated in high-dimensional systems. Recently some new ideas have been proposed to tackle this tricky problem. In this work, we use a newly developed method to resolve these inverse problems in different dynamical systems. Besides, we analyze the validity and precision of that theory and try to generalize this method. Finally, we try to apply this method to biological systems, e.g., how does a biological receptor extract the extracellular concentration of a signal molecule.