Fernando Morgado Dias

Learn More
Artificial Neural Networks became a common solution for a wide variety of problems in many fields, such as control and pattern recognition to name but a few. Many solutions found in these and other Artificial Neural Network fields have reached a hardware implementation phase, either commercial or with prototypes. The most frequent solution for the(More)
It is commonly assumed that neural networks have a built in fault tolerance property mainly due to their parallel structures. The international community of Neural Networks discussed these properties only until 1994 and afterwards the subject has been mostly ignored. Recently the subject was again brought to discussion due to the possibility of using neural(More)
Several implementations of Feedforward Neural Networks have been reported in scientific papers. These implementations do not allow the direct use of off-line trained networks. Usually the problem is the lower precision (compared to the software used for training) or modifications in the activation function. In the present work a hardware solution called(More)
This paper presents a technique for improving the fault tolerance capability of Artificial Neural Networks. This characteristic of distributed systems , which is usually pointed out as one of the advantages of this structure hasn't been deeply studied and can be improved in most of the networks. The solution implemented here consists of changing the(More)
The Levenberg-Marquardt algorithm is considered as the most effective one for training Artificial Neural Networks but its computational complexity and the difficulty to compute the trust region have made it very difficult to develop a true iterative version to use in on-line training. The algorithm is frequently used for off-line training in batch versions(More)
Artificial neural networks are a widespread tool with application in a variety of areas ranging from the social sciences to engineering. Many of these applications have reached a hardware implementation phase and have been documented in scientific papers. Unfortunately, most of the implementations have a simplified hyperbolic tangent replacement which has(More)
In the literature the most common proposed solutions for training inverse neural models are the direct (or general) and specialized methods. The second one being considered as more reliable to produce correct inverse models has nevertheless some drawbacks in the implementation. The present paper introduces a hybrid solution that copes with the problems and(More)
Natural neural networks greatly benefit from their parallel structure that makes them fault tolerant and fast in processing the inputs. Their artificial counterpart, artificial neural networks, proved difficult to implement in hardware where they could have a similar structure. Although, many circuits have been developed, they usually present problems(More)
This paper presents the FTSET tool for fault tolerance evaluation and improvement of Artificial Neural Networks. Fault tolerance is a characteristic of parallel distributed systems such as neural networks. Although there is a built-in fault tolerance in neural networks, it is possible to improve this characteristic, but changing the structure of an(More)