Leszek Rutkowski

Artificial Intelligence and Soft Computing

In recent years wind energy has been the fastest growing branch of the power generation industry. Maintenance of the wind turbine generates its the largest cost. A remote monitoring is a common method to reduce this cost. Growing number of monitored turbines requires an automatized way of support for diagnostic experts. Early fault detection and identification is still a very challenging task. A tool, which can alert an engineer about potentially dangerous cases, is required to work in real-time. The goal of this paper is to show an efficient system to online classification of operational states of the wind turbines and to detecting their early fault cases. The proposed system was designed as a hybrid of ART-2 and RBF networks. It had been proved before that the ART-type ANNs can successfully recognize operational states of a wind turbine during the diagnostic process. There are some difficulties, however, when classification is done in real-time. The disadvantages of using a classic ART-2 network are pointed and it is explained why the RBF unit of the hybrid system is needed to have a proper classification of turbine operational states.

Swarm and Evolutionary Computation

This paper presents a work inspired by the Pachycondyla apicalis ants behavior for the clustering problem. These ants have a simple but efficient prey search strategy: when they capture their prey, they return straight to their nest, drop off the prey and systematically return back to their original position. This behavior has already been applied to optimization, as the API meta-heuristic. API is a shortage of api-calis. Here, we combine API with the ability of ants to sort and cluster. We provide a comparison against Ant clustering Algorithm and K-Means using Machine Learning repository datasets. API introduces new concepts to ant-based models and gives us promising results.

Adaptive probabilistic neural networks for pattern classification in time-varying environment

In this paper, we propose a new class of probabilistic neural networks (PNNs) working in nonstationary environment. The novelty is summarized as follows: 1) We formulate the problem of pattern classification in nonstationary environment as the prediction problem and design a probabilistic neural network to classify patterns having time-varying probability distributions. We note that the problem of pattern classification in the nonstationary case is closely connected with the problem of prediction because on the basis of a learning sequence of the length n, a pattern in the moment n+k, k/spl ges/1 should be classified. 2) We present, for the first time in literature, definitions of optimality of PNNs in time-varying environment. Moreover, we prove that our PNNs asymptotically approach the Bayes-optimal (time-varying) decision surface. 3) We investigate the speed of convergence of constructed PNNs. 4) We design in detail PNNs based on Parzen kernels and multivariate Hermite series.

Decision Trees for Mining Data Streams Based on the McDiarmid's Bound

In mining data streams the most popular tool is the Hoeffding tree algorithm. It uses the Hoeffdings bound to determine the smallest number of examples needed at a node to select a splitting attribute. In the literature the same Hoeffdings bound was used for any evaluation function (heuristic measure), e.g., information gain or Gini index. In this paper, it is shown that the Hoeffdings inequality is not appropriate to solve the underlying problem. We prove two theorems presenting the McDiarmids bound for both the information gain, used in ID3 algorithm, and for Gini index, used in Classification and Regression Trees (CART) algorithm. The results of the paper guarantee that a decision tree learning system, applied to data streams and based on the McDiarmids bound, has the property that its output is nearly identical to that of a conventional learner. The results of the paper have a great impact on the state of the art of mining data streams and various developed so far methods and algorithms should be reconsidered.

Decision Trees for Mining Data Streams Based on the Gaussian Approximation

Since the Hoeffding tree algorithm was proposed in the literature, decision trees became one of the most popular tools for mining data streams. The key point of constructing the decision tree is to determine the best attribute to split the considered node. Several methods to solve this problem were presented so far. However, they are either wrongly mathematically justified (e.g., in the Hoeffding tree algorithm) or time-consuming (e.g., in the McDiarmid tree algorithm). In this paper, we propose a new method which significantly outperforms the McDiarmid tree algorithm and has a solid mathematical basis. Our method ensures, with a high probability set by the user, that the best attribute chosen in the considered node using a finite data sample is the same as it would be in the case of the whole data stream.

Neural Networks and Soft Computing

This paper discusses size-optimal solutions for implementing arbitrary Boolean functions using threshold gates. After presenting the state-of-the-art, we start from the result of Horne and Hush [12], which shows that threshold gate circuits restricted to fan-in 2 can implement arbitrary Boolean functions, but require O(2/n) gates in 2n layers. This result will be generalized to arbitrary fan-ins (∆), lowering the depth to n/log∆ + n/∆, and proving that all the (relative) minimums of size are obtained for sub-linear fan-ins (∆ < n − logn). The fact that size-optimal solutions have sub-linear fan-ins is encouraging, as the area and the delay of VLSI implementations are related to the fan-in of the gates.

The CART decision tree for mining data streams

One of the most popular tools for mining data streams are decision trees. In this paper we propose a new algorithm, which is based on the commonly known CART algorithm. The most important task in constructing decision trees for data streams is to determine the best attribute to make a split in the considered node. To solve this problem we apply the Gaussian approximation. The presented algorithm allows to obtain high accuracy of classification, with a short processing time. The main result of this paper is the theorem showing that the best attribute computed in considered node according to the available data sample is the same, with some high probability, as the attribute derived from the whole data stream.

Novel Online Speed Profile Generation for Industrial Machine Tool Based on Flexible Neuro-Fuzzy Approximation

Reference trajectory generation is one of the most important tasks in the control of machine tools. Such a trajectory must guarantee a smooth kinematics profile to avoid exciting the natural frequencies of the mechanical structure or servo control system. Moreover, the trajectory must be generated online to enable some feed rate adaptation mechanism working. This paper presents the online smooth speed profile generator used in trajectory interpolation in milling machines. Smooth kinematic profile is obtained by imposing limit on the jerk-which is the first derivative of acceleration. This generator is based on the neuro-fuzzy system and is able to adapt online the current feed rate to changing external conditions. Such an approach improves the machining quality, reduces the tool wear, and shortens total machining time. The proposed trajectory generation algorithm has been successfully tested and can be implemented on a multiaxis milling machine.

Fast image classification by boosting fuzzy classifiers

This paper presents a novel approach to visual objects classification based on generating simple fuzzy classifiers using local image features to distinguish between one known class and other classes. Boosting meta-learning is used to find the most representative local features. The proposed approach is tested on a state-of-the-art image dataset and compared with the bag-of-features image representation model combined with the Support Vector Machine classification. The novel method gives better classification accuracy and the time of learning and testing process is more than 30% shorter.

Connectionist Structures of Type 2 Fuzzy Inference Systems

In Fuzzy Inference Systems (FIS) the rule base consists of fuzzy relations between antecedents and consequents represented by classical fuzzy sets. Because their membership grades are exact real numbers in the unit interval [0, 1], there is no uncertainty in this sort of specification. In many applications there is some uncertainty as to the memberships, hence they can be stated as ordinary fuzzy sets of type 1 and can constitute type 2 fuzzy sets.In the world literature exists a global model of type 2 FIS. However it consists of an enormous number of embedded subsystems of type 1 and with regard to this model it has not found any use in connectionist realizations. In this paper we derive connectionist structures of type 2 FIS.