Hidden Markov models [2] (HMMs) have been successfully applied to Bioinformatics such as gene finding, remote homology detection and secondary structure prediction. On the other hand, continuous density HMMs have been widely used in the field of speech recognition. Though continuous density HMMs were not applied to Bioinformatics so far, they may also be useful in Bioinformatics. Currently, we ...

A fundamental challenge for any general theory of neural circuits is how to characterize the structure of the space of all possible circuits over a given model neuron. As a first step in this direction, this letter begins a systematic study of the global parameter space structure of continuous-time recurrent neural networks (CTRNNs), a class of neural models that is simple but dynamically unive...

During realistic, continuous perception, humans automatically segment experiences into discrete events. Using a novel model of cortical event dynamics, we investigate how cortical structures generate event representations during narrative perception and how these events are stored to and retrieved from memory. Our data-driven approach allows us to detect event boundaries as shifts between stabl...

FACULTY OF ENGINEERING, SCIENCE AND MATHEMATICS SCHOOL OF ELECTRONICS AND COMPUTER SCIENCE Doctor of Philosophy by Perukrishnen Vytelingum The last decade has seen a shift in emphasis from centralised to decentralised systems to meet the demanding coordination requirements of today’s complex computer systems. In such systems, the aim is to achieve effective decentralised control through autonom...

We study Bayesian networks for continuous variables using nonlinear conditional density estimators. We demonstrate that useful structures can be extracted from a data set in a self-organized way and we present sampling techniques for belief update based on Markov blanket conditional density models.

The growing number of digital audio and video repositories has resulted in a desperate need for eeective techniques to deliver data to users in a timely manner. Due to geographical distribution of users, it is not cost eeective to have a centralized media server. In this paper, we investigate issues involved in the design of a distributed video server (DVS) to support movie-on-demand (MOD) appl...

DR-submodular continuous functions are important objectives with wide real-world applications spanning MAP inference in determinantal point processes (DPPs), and mean-field inference for probabilistic submodular models, amongst others. DR-submodularity captures a subclass of non-convex functions that enables both exact minimization and approximate maximization in polynomial time. In this work w...

This paper is the continuation of [11] where the rotation construction of left-continuous triangular norms was presented. Here the class of triangular subnorms and a second construction, called rotation-annihilation, are introduced: Let T1 be a left-continuous triangular norm. If T1 has no zero divisors then let T2 be a left-continuous rotation invariant t-subnorm. If T1 has zero divisors then ...

This paper addresses the problem of choosing the multinomials that should compose a polynomial mathematical model starting from data. The mathematical representation used is a nonlinear differential equation of the polynomial type. Some approaches that have been used in the context of discrete-time models are adapted and applied to continuous-time models. Two examples are included to illustrate...