This paper studies the Generative Gaussian Graph proposed by Aupetit [1]. In particular, an automatic method for choosing the hyper-parameters in this algorithm is proposed. Experimental results are provided to demonstrate the performance of this proposed method.

This paper present a general method coupling genetic algorithms and Monte-Carlo simulation to address simulation optimization issues in the field of engineering asset management. After a description of the method, parameters tuning issues are analyzed through a test-case.

Minimalist grammars (MGs), as introduced in Stabler (1997), have proven a useful instrument in the formal analysis of syntactic theories developed within the minimalist branch of the principles–and–parameters framework (cf. Chomsky 1995, 2000). In fact, as shown in Michaelis (2001), MGs belong to the class of mildly context–sensitive grammars. Interestingly, without there being a rise in (at le...

Different works have shown that the combination of multiple loss functions is beneficial when training deep neural networks for a variety of prediction tasks. Generally, such multi-loss approaches are implemented via a weighted multi-loss objective function in which each term encodes a different desired inference criterion. The importance of each term is often set using empirically tuned hyper-...

This paper describes a small-scale, coarse-grained parallel Prolog-based Principles-and-Parameters parser that is derived essentially “for free”, modulo minor control flow changes, from an existing serial implementation. The grammar remains unchanged. The system is designed to operate as efficiently as possible under the severe constraints imposed by loosely-coupled processors. We demonstrate s...

The identification of kinetic models is an important step for the monitoring, control and optimization of chemical processes. Kinetic models are often based on first principles that describe the evolution of concentrations by means of conservation and constitutive equations. Identification of reaction kinetics, namely, rate expressions and rate parameters, represents the main challenge in const...

In many regards, tuning deep-learning networks is still more an art than it is a technique. Choosing the correct hyper-parameters and getting a complex network to learn properly can be daunting to people not well versed in that art. Taking the example of entailment classification, this work will analyze different methods and tools that can be used in order to diagnose learning problems. It will...

We present a method for generating surface crack patterns that appear in materials such as mud, ceramic glaze, and glass. To model these phenomena, we build upon existing physically based methods. Our algorithm generates cracks from a stress field defined heuristically over a triangle discretization of the surface. The simulation produces cracks by evolving this field over time. The user can co...

In the problem of parametric statistical inference with a nite parameter space, we study some simple rules for deening posterior upper and lower probabilities directly from the observed likelihood function, without using any prior probabilities. The rules satisfy the likelihood principle and a basic consistency principle (\avoiding sure loss"), they produce vacuous inferences when the likelihoo...

This paper describes a novel fuzzy rule-based modeling approach for some slow industrial processses. Structure identification is realized by clustering and support vector machines. When the process is slow, fuzzy rules can be obtained automatically. Parameters identification uses the techniques of fuzzy neural networks. A time-varying learning rate assures stability of the modeling error.