ARE ARTIFICIAL NEURAL NETS AS SMART AS A RAT

The origins of artificial neural networks are related to animal conditioning theory: both are forms of connectionist theory, which in turn derives from the empiricist philosophers' principle of association. The parallel between animal learning and neural nets suggests that interaction between them should benefit both sides.

Improved structure optimization for fuzzy-neural networks

Fuzzy-neural-network-based inference systems are well-known universal approximators which can produce linguistically interpretable results. Unfortunately, their dimensionality can be extremely high due to an excessive number of inputs and rules, which raises the need for overall structure optimization. In the literature, various input selection methods are available, but they are applied separately from rule selection, often without considering the fuzzy structure. This paper proposes an integrated framework to optimize the number of inputs and the number of rules simultaneously. First, a method is developed to select the most significant rules, along with a refinement stage to remove unnecessary correlations. An improved information criterion is then proposed to find an appropriate number of inputs and rules to include in the model, leading to a balanced tradeoff between interpretability and accuracy. Simulation results confirm the efficacy of the proposed method.

A New Gradient Descent Approach for Local Learning of Fuzzy Neural Models

The majority of reported learning methods for Takagi-Sugeno-Kang fuzzy neural models to date mainly focus on the improvement of their accuracy. However, one of the key design requirements in building an interpretable fuzzy model is that each obtained rule consequent must match well with the system local behaviour when all the rules are aggregated to produce the overall system output. This is one of the distinctive characteristics from black-box models such as neural networks. Therefore, how to find a desirable set of fuzzy partitions and, hence, to identify the corresponding consequent models which can be directly explained in terms of system behaviour presents a critical step in fuzzy neural modelling. In this paper, a new learning approach considering both nonlinear parameters in the rule premises and linear parameters in the rule consequents is proposed. Unlike the conventional two-stage optimization procedure widely practised in the field where the two sets of parameters are optimized separately, the consequent parameters are transformed into a dependent set on the premise parameters, thereby enabling the introduction of a new integrated gradient descent learning approach. A new Jacobian matrix is thus proposed and efficiently computed to achieve a more accurate approximation of the cost function by using the second-order Levenberg-Marquardt optimization method. Several other interpretability issues about the fuzzy neural model are also discussed and integrated into this new learning approach. Numerical examples are presented to illustrate the resultant structure of the fuzzy neural models and the effectiveness of the proposed new algorithm, and compared with the results from some well-known methods.

Determining propeller induced erosion alongside quay walls in harbours using Artificial Neural Networks

In shallow waters, such as those found close to berth structures, the wash from a manoeuvring ship’s propeller can cause erosion of the seabed. This erosion can be increased if the wash intersects a berth structure. A number of researchers have undertaken model studies and used regression analysis to develop predictive relationships for the scouring action. This paper presents an experimental investigation with Artificial Neural Networks (ANN’s), used to analyse the results. The purpose of using ANN’s was to examine the prediction accuracy of the Networks in comparison with previous regression analysis methods. ANN’s were found to provide a more accurate method of predicting propeller wash scour than the equations presented by previous investigators.

Trajectory Control of Electro Hydraulic Excavator using Fuzzy Self Tuning Algorithm with Neural Network

This paper presents the trajectory control of a 2DOF mini electro-hydraulic excavator by using fuzzy self tuning with neural network algorithm. First, the mathematical model is derived for the 2DOF mini electro-hydraulic excavator. The fuzzy PID and fuzzy self tuning with neural network are designed for circle trajectory following. Its two links are driven by an electric motor controlled pump system. The experimental results demonstrated that the proposed controllers have better control performance than the conventional controller.

Do different 'magnocellular tasks' probe the same neural substrate?

The sensory abnormalities associated with disorders such as dyslexia, autism and schizophrenia have often been attributed to a generalized deficit in the visual magnocellular-dorsal stream and its auditory homologue. To probe magnocellular function, various psychophysical tasks are often employed that require the processing of rapidly changing stimuli. But is performance on these several tasks supported by a common substrate? To answer this question, we tested a cohort of 1060 individuals on four 'magnocellular tasks': detection of low-spatial-frequency gratings reversing in contrast at a high temporal frequency (so-called frequency-doubled gratings); detection of pulsed low-spatial-frequency gratings on a steady luminance pedestal; detection of coherent motion; and auditory discrimination of temporal order. Although all tasks showed test-retest reliability, only one pair shared more than 4 per cent of variance. Correlations within the set of 'magnocellular tasks' were similar to the correlations between those tasks and a 'non-magnocellular task', and there was little consistency between 'magnocellular deficit' groups comprising individuals with the lowest sensitivity for each task. Our results suggest that different 'magnocellular tasks' reflect different sources of variance, and thus are not general measures of 'magnocellular function'.

Response of red blood cell folate to intervention: implications for folate recommendations for the prevention of neural tube defects

Committees worldwide have set almost identical folate recommendations for the prevention of the first occurrence of neural tube defects (NTDs). We evaluate these recommendations by reviewing the results of intervention studies that examined the response of red blood cell folate to altered folate intake. Three options are suggested to achieve the extra 400 mu g folic acid/d being recommended by the official committees: increased intake of folate-rich foods, dietary folic acid supplementation, and folic acid fortification of food. A significant increase in foods naturally rich in folates was shown to be a relatively ineffective means of increasing red blood cell folate status in women compared with equivalent intakes of folic acid-fortified food, presumably because the synthetic form of the vitamin is more stable and more bioavailable. Although folic acid supplements are highly effective in optimizing folate status, supplementation is not an effective strategy for the primary prevention of NTDs because of poor compliance. Thus, food fortification is seen by many as the only option likely to succeed. Mandatory folic acid fortification of grain products was introduced recently in the United States at a level projected to provide an additional mean intake of 100 mu g folic acid/d, but some feel that this policy does not go far enough. A recent clinical trial predicted that the additional intake of folic acid in the United States will reduce NTDs by >20%, whereas 200 mu g/d would be highly protective and is the dose also shown to be optimal in lowering plasma homocysteine, with possible benefits in preventing cardiovascular disease. Thus, an amount lower than the current target of an extra 400 mu g/d may be sufficient to increase red blood cell folate to concentrations associated with the lowest risk of NTDs, but further investigation is warranted to establish the optimal amount.

Fortification with low amounts of folic acid makes a significant difference in folate status in young women: implications for the prevention of neural tube defects

Background: Mandatory fortification of grain products with folic acid was introduced recently in the United States, a policy expected to result in a mean additional intake of 100 mu g/d. One way of predicting the effectiveness of this measure is to determine the effect of removing a similar amount of folic acid as fortified food from the diets of young women who had been electively exposed to chronic fortification.

Objective: The objective was to examine the effect on folate status of foods fortified with low amounts of folic acid.

Design: We investigated the changes in dietary intakes and in red blood cell and serum concentrations of folate in response to removing folic acid-fortified foods for 12 wk from the diets of women who reportedly consumed such foods at least once weekly (consumers).

Results: Consumers (n = 21) had higher total folate intakes (P = 0.002) and red blood cell folate concentrations (P = 0.023) than nonconsumers (women who consumed folic acid-fortified foods less than once weekly; n = 30). Of greater interest, a 12-wk intervention involving the exclusion of these foods resulted in a decrease in folate intake of 78 +/- 56 mu g/d (P < 0.001), which was reflected in a significant reduction in red blood cell folate concentrations (P < 0.05).

Conclusions: Cessation of eating folic acid-fortified foods resulted in removing 78 mu g folic acid/d from the diet. Over 12 wk this resulted in a lowering of red blood cell folate concentrations by 111 nmol/L (49 mu g/L). This magnitude of change in folate status in women can be anticipated as a result of the new US fortification legislation and is predicted to have a significant, although not optimal, effect in preventing neural tube defects.