Chong-Ho Choi

Input feature selection for classification problems

Feature selection plays an important role in classifying systems such as neural networks (NNs). We use a set of attributes which are relevant, irrelevant or redundant and from the viewpoint of managing a dataset which can be huge, reducing the number of attributes by selecting only the relevant ones is desirable. In doing so, higher performances with lower computational effort is expected. In this paper, we propose two feature selection algorithms. The limitation of mutual information feature selector (MIFS) is analyzed and a method to overcome this limitation is studied. One of the proposed algorithms makes more considered use of mutual information between input attributes and output classes than the MIFS. What is demonstrated is that the proposed method can provide the performance of the ideal greedy selection algorithm when information is distributed uniformly. The computational load for this algorithm is nearly the same as that of MIFS. In addition, another feature selection algorithm using the Taguchi method is proposed. This is advanced as a solution to the question as to how to identify good features with as few experiments as possible. The proposed algorithms are applied to several classification problems and compared with MIFS. These two algorithms can be combined to complement each others limitations. The combined algorithm performed well in several experiments and should prove to be a useful method in selecting features for classification problems.

Input feature selection by mutual information based on Parzen window

Mutual information is a good indicator of relevance between variables, and have been used as a measure in several feature selection algorithms. However, calculating the mutual information is difficult, and the performance of a feature selection algorithm depends on the accuracy of the mutual information. In this paper, we propose a new method of calculating mutual information between input and class variables based on the Parzen window, and we apply this to a feature selection algorithm for classification problems.

Iterative learning control in feedback systems

Abstract An iterative learning control method is proposed to achieve precise tracking control of a class of nonlinear systems over a finite time interval. The learning is done in a feedback configuration and the learning law updates the feedforward input from the plant input of the previous trial. A sufficient condition which guarantees the convergence of the learning is given. It is shown that the convergence condition of the learning control in the feedback configuration does not change from the condition in an open-loop configuration. But the learning speed can be improved greatly in the feedback configuration. Employing an input saturator which limits the control input within a reasonable bound, the class of nonlinear systems to which the proposed learning scheme can be applied is extended. The proposed learning control process is applied to the tracking control of a two link robot manipulator, and good tracking performance is obtained in the simulation.

Iterative learning control for a class of nonlinear systems

Abstract For precise tracking control of a class of nonlinear systems over a finite time interval, an iterative learning control method using the relative degree of a system is proposed. The class of nonlinear systems, where the iterative learning control is applied, can be extended by using the proposed method. A sufficient condition is derived for guaranteeing a uniform convergence of the output to the desired output. The results obtained for a class of nonlinear systems are shown to be a generalization of the existing results for linear time-invariant systems.

Dynamic compensation method for multivariable control systems with saturating actuators

This paper proposes a dynamic compensation method for multivariable control systems with saturating actuators to cope with the reset windup phenomenon. A dynamic compensator is explicitly determined based on a performance index of controller states. The proposed method is applicable to any open-loop stable plants with saturating actuators whose controller has been determined by some design technique. A simulation example is included to illustrate the effectiveness of the proposed method. >

Analysis and design of the virtual rate control algorithm for stabilizing queues in TCP networks

The virtual rate control (VRC) algorithm has been proposed for active queue management (AQM) in TCP networks. VRC, a rate-based control mechanism, responds quickly to traffic changes, thus allowing for high utilization and small loss. It can effectively stabilize both the input rate and the queue length around their target levels. In this paper, we analyze the stability of the VRC algorithm based on a linearized TCP model with time delay and provide a design guideline for parameter setting to make the overall system stable. Finally, we confirm the validity of our analysis and the effectiveness of VRC compared to RED, PI, REM, and AVQ through extensive ns-2 simulations.

Position control of X-Y table at velocity reversal using presliding friction characteristics

This paper aims at precision position control of the X-Y table of a computerized numeric control (CNC) machining center at velocity reversal. The characteristics of presliding friction are analyzed, and a simple and effective method is proposed to compensate the friction on the basis of these characteristics. A large position tracking error occurs at velocity reversal due to the sudden transition of friction between presliding regime and sliding regime. This paper investigates the transition time to reduce the tracking error, and derives a relationship between the transition time and the acceleration at zero velocity. This paper also proposes a method of estimating the transition time using this relationship, without having to measure velocity. Experimental observations confirm this correlation holds over a large dynamic range. In addition to friction, there is a large change in a torsional displacement at velocity reversal in a two-inertia system with finite stiffness like an X-Y table linked to a motor through a ballscrew. The proposed friction compensation scheme can be easily incorporated with the compensation method for torsional displacement to achieve good tracking performance. The experimental results are described to show the effectiveness of the proposed method.

Model-based disturbance attenuation for CNC machining centers in cutting process

A disturbance attenuation method in a control system, called the model-based disturbance attenuator (MBDA), is proposed, and its properties are studied. The MBDA makes the plant performs similarly to the nominal plant, as much as possible, using a compensator. Then, a controller is designed based on the nominal plant. It is shown that the MBDA is extremely robust with respect to large variations of load inertia. The MBDA is implemented in a position control system of a computer numerical control (CNC) machining center, where the velocity control system is composed of a servo-pack (PI controller), a servo motor, and a load. The MBDA attenuates external disturbances significantly in the cutting process containing high-frequency components, as well as the frictional forces containing large DC component. Several other controllers are also implemented in a position control system of a CNC machining center in a similar way as the MBDA, and the experimental results are compared with one another.

Feature extraction based on ICA for binary classification problems

In manipulating data such as in supervised learning, we often extract new features from the original features for the purpose of reducing the dimensions of feature space and achieving better performance. In this paper, we show how standard algorithms for independent component analysis (ICA) can be appended with binary class labels to produce a number of features that do not carry information about the class labels-these features will be discarded-and a number of features that do. We also provide a local stability analysis of the proposed algorithm. The advantage is that general ICA algorithms become available to a task of feature extraction for classification problems by maximizing the joint mutual information between class labels and new features, although only for two-class problems. Using the new features, we can greatly reduce the dimension of feature space without degrading the performance of classifying systems.

Shadow compensation in 2D images for face recognition

Illumination variation that occurs on face images degrades the performance of face recognition. In this paper, we propose a novel approach to handling illumination variation for face recognition. Since most human faces are similar in shape, we can find the shadow characteristics, which the illumination variation makes on the faces depending on the direction of light. By using these characteristics, we can compensate for the illumination variation on face images. The proposed method is simple and requires much less computational effort than the other methods based on 3D models, and at the same time, provides a comparable recognition rate.