Comparing feedforward neural networks using independent component analysis on hidden units.

Neural networks are widely used for classification and regression tasks, but they do not always perform well, nor explicitly inform us of the rationale for their predictions. In this study we propose a novel method of comparing a pair of different feedforward neural networks, which draws on independent components obtained by independent component analysis (ICA) on the hidden layers of these networks. It can compare different feedforward neural networks even when they have different structures, as well as feedforward neural networks that learned partially different datasets, yielding insights into their functionality or performance. We evaluate the proposed method by conducting three experiments with feedforward neural networks that have one hidden layer, and verify whether a pair of feedforward neural networks can be compared by the proposed method when the numbers of hidden units in the layer are different, when the datasets are partially different, and when activation functions are different. The results show that similar independent components are extracted from two feedforward neural networks, even when the three circumstances above are different. Our experiments also reveal that mere comparison of weights or activations does not lead to identifying similar relationships. Through the extraction of independent components, the proposed method can assess whether the internal processing of one neural network resembles that of another. This approach has the potential to help understand the performance of neural networks.

Clinical outcome and mechanism of soft tissue calcification in Werner syndrome.

Werner syndrome (WS) is an autosomal recessive progeroid disorder caused by mutations in RecQ DNA helicase. Ectopic soft tissue calcification is one of the well known symptoms in WS. However, the prevalence, clinical outcome, and mechanism of such calcification remain to be elucidated. The clinical features and mechanism of ectopic calcification were examined in seven patients with WS whose diagnosis were confirmed by a genomic DNA analysis. X-ray examinations revealed subcutaneous calcification in 35 of 41 major joints (85.3%). The patients complained of dermal pain at 23 joints among 35 joints (65.7%) with calcification. Refractory skin ulcers were found at the area of the skin overlaying the calcification in 16 joints (45.7%). In contrast, no pain or ulcers were observed in the joints without calcification. The presence of ectopic calcification could not be explained by a systemic hormonal abnormality. Cultured fibroblasts from WS patients underwent spontaneous mineralization in vitro in the normal phosphate condition, and overexpressed Pit-1, a transmembrane type III Na-Pi cotransporter both at the mRNA and protein levels. Phosphonophormic acid, a specific inhibitor for Pit-1, inhibited mineralization in the WS fibroblasts. Both calcification and Pit-1 overexpression were detected in the skin of WS in situ. WS showed a high prevalence of ectopic calcification, which was associated with dermal pain and refractory skin ulcers. An overexpression of Pit-1 therefore seems to play a key role in the formation of soft tissue calcification in this syndrome.