We consider feed-forward neural networks with one non-linear hidden layer and linear output units. The transfer function in the hidden layer are either bell-shaped or sigmoid. In the bell-shaped case, we show how Bernstein polynomials on one hand and the theory of the heat equation on the other are relevant for understanding the properties of the corresponding networks. In particular, these tec...

The aim of this work was to develop a pulsatile release system with metoprolol for chronotherapeutical use by coating swellable mini-tablets with Eudragit RS. To study the influence of the formulation factors (amount of coating polymer, plasticizer percentage in film coating and swelling agent percentage in mini-tablets), a Box-Behnken design of experiment (DoE) was used. To evaluate the influe...

In this paper, a revised Group Method of Data Handling (GMDH)-type neural network algorithm with a feedback loop identifying sigmoid function neural network is proposed. In this algorithm, the optimum sigmoid function neural network architecture is automatically organized so as to minimize the prediction error criterion defined as Akaike’s Information Criterion (AIC) or Prediction Sum of Square...

BACKGROUND The feasibility of laparoscopic sigmoid colectomy for diverticular disease has now been well established. We report herein our experience with laparoscopic sigmoid colectomy in 100 patients who underwent laparoscopic colectomy for chronic diverticular disease. METHODS A retrospective review was performed of a 7-year period from January 1995 to June 2002. Chronic diverticular diseas...

We present deterministic techniques for com­ puting upper and lower bounds on marginal probabilities in sigmoid and noisy-OR net­ works. These techniques become useful when the size of the network (or clique size) pre­ cludes exact computations. We illustrate the tightness of the bounds by numerical experi­ ments.

The Squeeze-and-Excitation (SE) structure has been designed to enhance the neural network performance by allowing it execute positive channel-wise feature recalibration and suppress less useful features. SE structures are generally adopted in a plethora of tasks directly existing models have shown actual enhancements. However, various sigmoid functions used artificial networks intrinsically res...

In this paper, we revise two commonly used saturated functions, the logistic sigmoid and the hyperbolic tangent (tanh). We point out that, besides the well-known non-zero centered property, slope of the activation function near the origin is another possible reason making training deep networks with the logistic function difficult to train. We demonstrate that, with proper rescaling, the logist...

|The performance of real-time machine vision for workpiece manipulation is postulated to be describable by a sigmoid function of the computational resources made available for this purpose. Several attempts are discussed to estimate a representative resource reference value which characterizes the argument range where the hypothesized sigmoid function exhibits its steepest slope. Provided compu...

Activation functions have a notorious impact on neural networks on both training and testing the models against the desired problem. Currently, the most used activation function is the Rectified Linear Unit (ReLU). This paper introduces a new and novel activation function, closely related with the new activation Swish = x ∗ sigmoid(x) (Ramachandran et al., 2017) [14] which generalizes it. We ca...

Congenital prosopagnosia (cPA) is a severe disorder in recognising familiar faces, a human characteristic that is presumably innate, without any macro-spatial brain anomalies. Following the idea that cPA is based on deficits of configural face processing, we used a speeded grotesqueness decision task with thatcherised faces, since the Thatcher illusion can serve as a test of configural disrupti...