Sparse kernel density construction using orthogonal forward regression with leave-one-out test score and local regularization

This paper presents an efficient construction algorithm for obtaining sparse kernel density estimates based on a regression approach that directly optimizes model generalization capability. Computational efficiency of the density construction is ensured using an orthogonal forward regression, and the algorithm incrementally minimizes the leave-one-out test score. A local regularization method is incorporated naturally into the density construction process to further enforce sparsity. An additional advantage of the proposed algorithm is that it is fully automatic and the user is not required to specify any criterion to terminate the density construction procedure. This is in contrast to an existing state-of-art kernel density estimation method using the support vector machine (SVM), where the user is required to specify some critical algorithm parameter. Several examples are included to demonstrate the ability of the proposed algorithm to effectively construct a very sparse kernel density estimate with comparable accuracy to that of the full sample optimized Parzen window density estimate. Our experimental results also demonstrate that the proposed algorithm compares favorably with the SVM method, in terms of both test accuracy and sparsity, for constructing kernel density estimates.

A kernel-based two-class classifier for imbalanced data sets

Many kernel classifier construction algorithms adopt classification accuracy as performance metrics in model evaluation. Moreover, equal weighting is often applied to each data sample in parameter estimation. These modeling practices often become problematic if the data sets are imbalanced. We present a kernel classifier construction algorithm using orthogonal forward selection (OFS) in order to optimize the model generalization for imbalanced two-class data sets. This kernel classifier identification algorithm is based on a new regularized orthogonal weighted least squares (ROWLS) estimator and the model selection criterion of maximal leave-one-out area under curve (LOO-AUC) of the receiver operating characteristics (ROCs). It is shown that, owing to the orthogonalization procedure, the LOO-AUC can be calculated via an analytic formula based on the new regularized orthogonal weighted least squares parameter estimator, without actually splitting the estimation data set. The proposed algorithm can achieve minimal computational expense via a set of forward recursive updating formula in searching model terms with maximal incremental LOO-AUC value. Numerical examples are used to demonstrate the efficacy of the algorithm.

A forward-constrained regression algorithm for sparse kernel density estimation

Using the classical Parzen window (PW) estimate as the target function, the sparse kernel density estimator is constructed in a forward-constrained regression (FCR) manner. The proposed algorithm selects significant kernels one at a time, while the leave-one-out (LOO) test score is minimized subject to a simple positivity constraint in each forward stage. The model parameter estimation in each forward stage is simply the solution of jackknife parameter estimator for a single parameter, subject to the same positivity constraint check. For each selected kernels, the associated kernel width is updated via the Gauss-Newton method with the model parameter estimate fixed. The proposed approach is simple to implement and the associated computational cost is very low. Numerical examples are employed to demonstrate the efficacy of the proposed approach.

Model selection approaches for non-linear system identification: a review

The identification of non-linear systems using only observed finite datasets has become a mature research area over the last two decades. A class of linear-in-the-parameter models with universal approximation capabilities have been intensively studied and widely used due to the availability of many linear-learning algorithms and their inherent convergence conditions. This article presents a systematic overview of basic research on model selection approaches for linear-in-the-parameter models. One of the fundamental problems in non-linear system identification is to find the minimal model with the best model generalisation performance from observational data only. The important concepts in achieving good model generalisation used in various non-linear system-identification algorithms are first reviewed, including Bayesian parameter regularisation and models selective criteria based on the cross validation and experimental design. A significant advance in machine learning has been the development of the support vector machine as a means for identifying kernel models based on the structural risk minimisation principle. The developments on the convex optimisation-based model construction algorithms including the support vector regression algorithms are outlined. Input selection algorithms and on-line system identification algorithms are also included in this review. Finally, some industrial applications of non-linear models are discussed.

Semantic-associative visual content labelling and retrieval: A multimodal approach

A novel framework referred to as collaterally confirmed labelling (CCL) is proposed, aiming at localising the visual semantics to regions of interest in images with textual keywords. Both the primary image and collateral textual modalities are exploited in a mutually co-referencing and complementary fashion. The collateral content and context-based knowledge is used to bias the mapping from the low-level region-based visual primitives to the high-level visual concepts defined in a visual vocabulary. We introduce the notion of collateral context, which is represented as a co-occurrence matrix of the visual keywords. A collaborative mapping scheme is devised using statistical methods like Gaussian distribution or Euclidean distance together with collateral content and context-driven inference mechanism. We introduce a novel high-level visual content descriptor that is devised for performing semantic-based image classification and retrieval. The proposed image feature vector model is fundamentally underpinned by the CCL framework. Two different high-level image feature vector models are developed based on the CCL labelling of results for the purposes of image data clustering and retrieval, respectively. A subset of the Corel image collection has been used for evaluating our proposed method. The experimental results to-date already indicate that the proposed semantic-based visual content descriptors outperform both traditional visual and textual image feature models. (C) 2007 Elsevier B.V. All rights reserved.

Collaterally cued labelling framework underpinning semantic-level visual content descriptor

In this paper, we introduce a novel high-level visual content descriptor devised for performing semantic-based image classification and retrieval. The work can be treated as an attempt for bridging the so called "semantic gap". The proposed image feature vector model is fundamentally underpinned by an automatic image labelling framework, called Collaterally Cued Labelling (CCL), which incorporates the collateral knowledge extracted from the collateral texts accompanying the images with the state-of-the-art low-level visual feature extraction techniques for automatically assigning textual keywords to image regions. A subset of the Corel image collection was used for evaluating the proposed method. The experimental results indicate that our semantic-level visual content descriptors outperform both conventional visual and textual image feature models.

Sparse kernel density estimation technique based on zero-norm constraint

A sparse kernel density estimator is derived based on the zero-norm constraint, in which the zero-norm of the kernel weights is incorporated to enhance model sparsity. The classical Parzen window estimate is adopted as the desired response for density estimation, and an approximate function of the zero-norm is used for achieving mathemtical tractability and algorithmic efficiency. Under the mild condition of the positive definite design matrix, the kernel weights of the proposed density estimator based on the zero-norm approximation can be obtained using the multiplicative nonnegative quadratic programming algorithm. Using the -optimality based selection algorithm as the preprocessing to select a small significant subset design matrix, the proposed zero-norm based approach offers an effective means for constructing very sparse kernel density estimates with excellent generalisation performance.

Lexical access across languages: a multinomial model of auditory distraction

Recall in many types of verbal memory task is reliably disrupted by the presence of auditory distracters, with verbal distracters frequently proving the most disruptive (Beaman, 2005). A multinomial processing tree model (Schweickert, 1993) is applied to the effects on free recall of background speech from a known or an unknown language. The model reproduces the free recall curve and the impact on memory of verbal distracters for which a lexical entry exists (i.e., verbal items from a known language). The effects of semantic relatedness of distracters within a language is found to depend upon a redintegrative factor thought to reflect the contribution of the speech-production system. The differential impacts of known and unknown languages cannot be accounted for in this way, but the same effects of distraction are observed amongst bilinguals, regardless of distracter-language.

Wernicke’s aphasia reflects a combination of acoustic-phonological and semantic control deficits: A case-series comparison of Wernicke’s aphasia, semantic dementia and semantic aphasia

Wernicke’s aphasia (WA) is the classical neurological model of comprehension impairment and, as a result, the posterior temporal lobe is assumed to be critical to semantic cognition. This conclusion is potentially confused by (a) the existence of patient groups with semantic impairment following damage to other brain regions (semantic dementia and semantic aphasia) and (b) an ongoing debate about the underlying causes of comprehension impairment in WA. By directly comparing these three patient groups for the first time, we demonstrate that the comprehension impairment in Wernicke’s aphasia is best accounted for by dual deficits in acoustic-phonological analysis (associated with pSTG) and semantic cognition (associated with pMTG and angular gyrus). The WA group were impaired on both nonverbal and verbal comprehension assessments consistent with a generalised semantic impairment. This semantic deficit was most similar in nature to that of the semantic aphasia group suggestive of a disruption to semantic control processes. In addition, only the WA group showed a strong effect of input modality on comprehension, with accuracy decreasing considerably as acoustic-phonological requirements increased. These results deviate from traditional accounts which emphasise a single impairment and, instead, implicate two deficits underlying the comprehension disorder in WA.

Towards the design of a unified semantics model for OLTP and OLAP systems

The concepts of on-line transactional processing (OLTP) and on-line analytical processing (OLAP) are often confused with the technologies or models that are used to design transactional and analytics based information systems. This in some way has contributed to existence of gaps between the semantics in information captured during transactional processing and information stored for analytical use. In this paper, we propose the use of a unified semantics design model, as a solution to help bridge the semantic gaps between data captured by OLTP systems and the information provided by OLAP systems. The central focus of this design approach is on enabling business intelligence using not just data, but data with context.

L3 syntactic transfer selectivity and typological determinacy: the typological primacy model

The present article addresses the following question: what variables condition syntactic transfer? Evidence is provided in support of the position that third language (L3) transfer is selective, whereby, at least under certain conditions, it is driven by the typological proximity of the target L3 measured against the other previously acquired linguistic systems (cf. Rothman and Cabrelli Amaro, 2007, 2010; Rothman, 2010; Montrul et al., 2011). To show this, we compare data in the domain of adjectival interpretation between successful first language (L1) Italian learners of English as a second language (L2) at the low to intermediate proficiency level of L3 Spanish, and successful L1 English learners of L2 Spanish at the same levels for L3 Brazilian Portuguese. The data show that, irrespective of the L1 or the L2, these L3 learners demonstrate target knowledge of subtle adjectival semantic nuances obtained via noun-raising, which English lacks and the other languages share. We maintain that such knowledge is transferred to the L3 from Italian (L1) and Spanish (L2) respectively in light of important differences between the L3 learners herein compared to what is known of the L2 Spanish performance of L1 English speakers at the same level of proficiency (see, for example, Judy et al., 2008; Rothman et al., 2010). While the present data are consistent with Flynn et al.’s (2004) Cumulative Enhancement Model, we discuss why a coupling of these data with evidence from other recent L3 studies suggests necessary modifications to this model, offering in its stead the Typological Primacy Model (TPM) for multilingual transfer.

Does the PMSP connectionist model of single word reading learn to read in the same way as a child?

The Plaut, McClelland, Seidenberg and Patterson (1996) connectionist model of reading was evaluated at two points early in its training against reading data collected from British children on two occasions during their first year of literacy instruction. First, the network’s non-word reading was poor relative to word reading when compared with the children. Second, the network made more non-lexical than lexical errors, the opposite pattern to the children. Three adaptations were made to the training of the network to bring it closer to the learning environment of a child: an incremental training regime was adopted; the network was trained on grapheme– phoneme correspondences; and a training corpus based on words found in children’s early reading materials was used. The modifications caused a sharp improvement in non-word reading, relative to word reading, resulting in a near perfect match to the children’s data on this measure. The modified network, however, continued to make predominantly non-lexical errors, although evidence from a small-scale implementation of the full triangle framework suggests that this limitation stems from the lack of a semantic pathway. Taken together, these results suggest that, when properly trained, connectionist models of word reading can offer insights into key aspects of reading development in children.

Fast identification algorithms for Gaussian process model

A class identification algorithms is introduced for Gaussian process(GP)models.The fundamental approach is to propose a new kernel function which leads to a covariance matrix with low rank,a property that is consequently exploited for computational efficiency for both model parameter estimation and model predictions.The objective of either maximizing the marginal likelihood or the Kullback–Leibler (K–L) divergence between the estimated output probability density function(pdf)and the true pdf has been used as respective cost functions.For each cost function,an efficient coordinate descent algorithm is proposed to estimate the kernel parameters using a one dimensional derivative free search, and noise variance using a fast gradient descent algorithm. Numerical examples are included to demonstrate the effectiveness of the new identification approaches.

Estimation of Gaussian process regression model using probability distance measures

A new class of parameter estimation algorithms is introduced for Gaussian process regression (GPR) models. It is shown that the integration of the GPR model with probability distance measures of (i) the integrated square error and (ii) Kullback–Leibler (K–L) divergence are analytically tractable. An efficient coordinate descent algorithm is proposed to iteratively estimate the kernel width using golden section search which includes a fast gradient descent algorithm as an inner loop to estimate the noise variance. Numerical examples are included to demonstrate the effectiveness of the new identification approaches.

Meeting detection in video through semantic analysis

In this paper we present a novel approach to detect people meeting. The proposed approach works by translating people behaviour from trajectory information into semantic terms. Having available a semantic model of the meeting behaviour, the event detection is performed in the semantic domain. The model is learnt employing a soft-computing clustering algorithm that combines trajectory information and motion semantic terms. A stable representation can be obtained from a series of examples. Results obtained on a series of videos with different types of meeting situations show that the proposed approach can learn a generic model that can effectively be applied on the behaviour recognition of meeting situations.