A node pruning algorithm based on a Fourier amplitude sensitivity test method.

In this paper, we propose a new pruning algorithm to obtain the optimal number of hidden units of a single layer of a fully connected neural network (NN). The technique relies on a global sensitivity analysis of model output. The relevance of the hidden nodes is determined by analysing the Fourier decomposition of the variance of the model output. Each hidden unit is assigned a ratio (the fraction of variance which the unit accounts for) that gives their ranking. This quantitative information therefore leads to a suggestion of the most favorable units to eliminate. Experimental results suggest that the method can be seen as an effective tool available to the user in controlling the complexity in NNs.

Surgical margins in cutaneous melanoma (2 cm versus 5 cm for lesions measuring less than 2.1-mm thick).

BACKGROUND: This study addressed the question of whether limited surgery for primary malignant melanoma with a 2-cm margin is as good as a 5-cm margin. An update of a 16-year follow-up is provided. METHODS: Nine European Centers, over a period of 5 years, prospectively randomized 337 patients with melanoma measuring less than 2.1 mm in thickness to undergo a local excision with either a 2-cm or a 5-cm margin. Three hundred twenty-six patients were eligible for statistical analysis. Excluded from the trial were patients older than 70 years; those with melanomas from the toe, nail, or finger; and those with acral-lentiginous melanoma. A separate randomization was performed to independently test an adjuvant treatment with a nonspecific immunostimulant, isoprinosine, compared with observation. The median follow-up time was 192 months (16 years) for the estimation of survival and disease recurrences. RESULTS: There were 22 tumor recurrences in the 2-cm arm and 33 in the 5-cm arm. The median time to disease recurrence was 43 months and 37.6 months, respectively. The 10-year disease-free survival rates were 85% for the group with a 2-cm margin and 83% for the group with a 5-cm margin. There was no difference in the 10-year overall survival rates (87% vs. 86%). Isoprinosine did not demonstrate any activity in this setting. CONCLUSIONS: The authors concluded that for melanoma less than 2.1-mm thick, a margin of excision of 2 cm is sufficient. A larger margin of 5 cm does not appear to have any impact on either the rate or the time to disease recurrence or on survival.