Joint training of non-negative Tucker decomposition and discrete density hidden Markov models

Non-negative Tucker decomposition (NTD) is applied to unsupervised training of discrete density HMMs for the discovery of sequential patterns in data, for segmenting sequential data into patterns and for recognition of the discovered patterns in unseen data. Structure constraints are imposed on the NTD such that it shares its parameters with the HMM. Two training schemes are proposed: one uses NTD as a regularizer for the Baum–Welch (BW) training of the HMM, the other alternates between initializing the NTD with the BW output and vice versa. On the task of unsupervised spoken pattern discovery from the TIDIGITS database, both training schemes are observed to improve over BW training in terms of pattern purity, accuracy of the segmentation boundaries and accuracy for speech recognition. Furthermore, we experimentally observe that the alternative training of NTD and BW outperforms the NTD regularized BW, BW training and BW training with simulated annealing.     

SIFER: Scale-Invariant Feature Detector with Error Resilience

We present a new method to extract scale-invariant features from an image by using a Cosine Modulated Gaussian (CM-Gaussian) filter. Its balanced scale-space atom with minimal spread in scale and space leads to an outstanding scale-invariant feature detection quality, albeit at reduced planar rotational invariance. Both sharp and distributed features like corners and blobs are reliably detected, irrespective of various image artifacts and camera parameter variations, except for planar rotation. The CM-Gaussian filters are approximated with the sum of exponentials as a single, fixed-length filter and equal approximation error over all scales, providing constant-time, low-cost image filtering implementations. The approximation error of the corresponding digital signal processing is below the noise threshold. It is scalable with the filter order, providing many quality-complexity trade-off working points. We validate the efficiency of the proposed feature detection algorithm on image registration applications over a wide range of testbench conditions.     

Training Effective Node Classifiers for Cascade Classification

Cascade classifiers are widely used in real-time object detection. Different from conventional classifiers that are designed for a low overall classification error rate, a classifier in each node of the cascade is required to achieve an extremely high detection rate and moderate false positive rate. Although there are a few reported methods addressing this requirement in the context of object detection, there is no principled feature selection method that explicitly takes into account this asymmetric node learning objective. We provide such an algorithm here. We show that a special case of the biased minimax probability machine has the same formulation as the linear asymmetric classifier (LAC) of Wu et al. (linear asymmetric classifier for cascade detectors, 2005). We then design a new boosting algorithm that directly optimizes the cost function of LAC. The resulting totally-corrective boosting algorithm is implemented by the column generation technique in convex optimization. Experimental results on object detection verify the effectiveness of the proposed boosting algorithm as a node classifier in cascade object detection, and show performance better than that of the current state-of-the-art.     

Time series classification for the prediction of dialysis in critically ill patients using echo statenetworks

Objective: Time series often appear in medical databases, but only few machine learning methods exist that process this kind of data properly. Most modeling techniques have been designed with a static data model in mind and are not suitable for coping with the dynamic nature of time series. Recurrent neural networks (RNNs) are often used to process time series, but only a few training algorithms exist for RNNs which are complex and often yield poor results. Therefore, researchers often turn to traditional machine learning approaches, such as support vector machines (SVMs), which can easily be set up and trained and combine them with feature extraction (FE) and selection (FS) to process the high-dimensional temporal data. Recently, a new approach, called echo state networks (ESNs), has been developed to simplify the training process of RNNs. This approach allows modeling the dynamics of a system based on time series data in a straightforwardway.The objective of this study is to explore the advantages of using ESN instead of other traditional classifiers combined with FE and FS in classification problems in the intensive care unit (ICU) when the input data consists of time series. While ESNs have mostly been used to predict the future course of a time series, we use the ESN model for classification instead. Although time series often appear in medical data, little medical applications of ESNs have been studiedyet.Methods and material: ESN is used to predict the need for dialysis between the fifth and tenth day after admission in the ICU. The input time series consist of measured diuresis and creatinine values during the first 3days after admission. Data about 830 patients was used for the study, of which 82 needed dialysis between the fifth and tenth day after admission. ESN is compared to traditional classifiers, a sophisticated and a simple one, namely support vector machines and the naive Bayes (NB) classifier. Prior to the use of the SVM and NB classifier, FE and FS is required to reduce the number of input features and thus alleviate the curse dimensionality. Extensive feature extraction was applied to capture both the overall properties of the time series and the correlation between the different measurements in the time series. The feature selection method consists of a greedy hybrid filter-wrapper method using a NB classifier, which selects in each iteration the feature that improves prediction the best and shows little multicollinearity with the already selected set. Least squares regression with noise was used to train the linear readout function of the ESN to mitigate sensitivity to noise and overfitting. Fisher labeling was used to deal with the unbalanced data set. Parameter sweeps were performed to determine the optimal parameter values for the different classifiers. The area under the curve (AUC) and maximum balanced accuracy are used as performance measures. The required execution time was also measured.Results: The classification performance of the ESN shows significant difference at the 5% level compared to the performance of the SVM or the NB classifier combined with FE and FS. The NB+FE+FS, with an average AUC of 0.874, has the best classification performance. This classifier is followed by the ESN, which has an average AUC of 0.849. The SVM+FE+FS has the worst performance with an average AUC of 0.838. The computation time needed to pre-process the data and to train and test the classifier is significantly less for the ESN compared to the SVM andNB.Conclusion: It can be concluded that the use of ESN has an added value in predicting the need for dialysis through the analysis of time series data. The ESN requires significantly less processing time, needs no domain knowledge, is easy to implement, and can be configured using rules ofthumb.     

Lyapunov based predictive control of vehicle drivetrains over CAN

Vehicle drivetrains are characterized by fast dynamics, subject to physical and control constraints, which make controller design for driveline oscillations damping a challenging problem. Furthermore, in current implementations, the connections between the controller and the physical plant are realized using a controller area network (CAN) as the communication medium, which introduces time-varying delays. As such, the goal of this paper is to provide a control design methodology that can cope with all these challenges and limitations and still yield an effective solution. To this end, firstly, a continuous-time model of a vehicle drivetrain is derived. Then, a method for determining a worst case upper bound on the delays that can be introduced by a CAN is presented, which enables the usage of a polytopic approximation technique to obtain a discrete-time model of the closed-loop CAN system. Thirdly, a non-conservative Lyapunov based predictive controller is designed for the resulting model with time-varying delays, polytopic uncertainty and hard constraints. Several tests performed using an industry validated drivetrain model and the Matlab toolbox TrueTime indicate that the proposed design methodology can handle both the performance/physical constraints and the strict limitations on the computational complexity, while effectively coping with time-varying delays. Preliminary real-time results further validate the proposed methodology.     

A hybrid approach to modelling,control and state estimation of mechanical systems with backlash

Control of mechanical systems with backlash is a topic well studied by many control practitioners. This interest has been motivated by the fact that backlash in mechanical systems can cause severe performance degradation and lead to instability of the control system. Furthermore, high impact-forces in backlash-systems can lead to a lower durability of the components and to strokes and peaks in the output. In this paper a mechanical benchmark system is presented to provide facilities for testing the identification and control of systems with backlash. For controller design a hybrid model of the system was derived and used in a model predictive control (MPC) scheme. Observer-based state-estimation was used to recover unmeasured states, particularly the backlash angle. Explicit solutions of a tracking controller were computed to control the mechanical benchmark system in real-time. Simulation as well as experimental results are presented to show the applicability of this hybrid control approach.     

Design of an analytic constrained predictive controller using neural networks

This paper shows how the solution of the standard predictive control problem can be recast as a continuous function of the state, the reference signal, the noise and the disturbances, and hence can be approximated arbitrarily closely by a feed-forward neural network. The existence of such a continuous mapping eliminates the need for linear independency of the active constraints, and therefore the resulting analytic constrained predictive controller will combine constraint handling with speed while being applicable to fast and complex control systems with many constraints. The effectiveness of the proposed controller design methodology is shown for a simulation example of an elevator model and for a real-time laboratory inverted pendulum system.     

Robust H ∞ filtering for a class of non-linear systems with state delay and parameter uncertainty

This paper deals with the problem of robust H X filtering for a class of state-delayed non-linear systems with normbounded parameter uncertainty appearing in all the matrices of the linear part of the system model. The non-linearities are assumed to satisfy the global Lipschitz conditions and appear in both the state and measured output equations. Attention is focused on the design of a non-linear filter which ensures both the robust stability and a prescribed H X performance of the filtering error dynamics for all admissible uncertainties. A sufficient condition for the existence of such a filter is given in terms of a linear matrix inequality (LMI). When this LMI is feasible, the expression of a desired H X filter is also presented. A numerical example is provided to demonstrate the applicability of the proposed approach.     

Data fusion for investigating land subsidence and coal fire hazards in a coal mining area

Coal mining areas all over the world are often threatened by serious environmental hazards such as the occurrence of coal fires, land subsidence, etc. Coal fires burn away the natural non-renewable coal resources, locally raise the temperature of the area, emit polluting gases such as oxides of carbon, sulphur and nitrogen, and when present underground are even the cause of land subsidence. Mining-induced subsidences, on the other hand, cause horizontal and vertical movements in the land surface, and open cracks and fissures that serve as inlets for oxygen, which in turn aggravate the problem of coal fires. These inter-related phenomena often render the mining areas unfit for human inhabitation and the commercial exploitation of coal nearly impossible in some parts. In this study, satellite data acquired in three regions of the electromagnetic spectrum, namely optical, thermal and microwave, along with field data, are used to identify the areas affected by coal fires and land subsidence in a coalfield in north-west China. Data fusion techniques are used for an integrated analysis of this complex problem.