A Novel Clustering Method Combining Heuristics and Information Theorem

Despite of its simplicity and success in various applications, conventional competitive learning (CL) making use of the winner-take-all strategy suffers from two major shortcomings, i.e. neuron under utilization and waste of closeness information computed. In this paper, a fuzzy approach to address these shortcomings is pursued. By considering the concept "win" as a fuzzy set, two existing competitive learning algorithms, namely the standard CL algorithm and the frequency sensitive CL algorithm, are generalized and the resulting fuzzy algorithms are proposed. Furthermore, a monotonically decreasing implementation scheme for the fuzziness parameter introduced in the proposed algorithms is suggested to further enhance the overall performance of the fuzzy algorithms. The effectiveness of the proposed algorithms is demonstrated with numerical examples.

For the classical k-means clustering algorithm, the problem of selecting an appropriate k is a hard problem and affects the performance of k-means strongly. When used for clustering analysis, the conventional competitive learning (CL) algorithms also have a similar crucial problem-the selection of an appropriate number of neural units. The performance of frequency sensitive competitive learning (FSCL)-one version of the improved CL algorithms, also significantly deteriorates when the number of units is inappropriately selected. The paper proposes a new algorithm called rival penalized competitive learning (RPCL), which has the ability of automatically allocating an appropriate number of units for an input data set. The experimental results have shown that RPCL outperforms FSCL significantly when they are used for unsupervised classification, and supervised classification through the radial basis function net. >

The dynamic competitive learning method

DECO3R: A Differential Evolution-based algorithm for generating compact Fuzzy Rule-based Classification Systems

In 3D computer graphics, mesh parameterization is a key technique for digital geometry processings such as morphing, shape blending, texture mapping, re-meshing and so on. Most of the previous approaches made use of an identical primitive domain to parameterize a mesh model. In recent works of mesh parameterization, more flexible and attractive methods that can create direct mappings between two meshes have been reported. These mappings are called “cross-parameterization” and typically preserve semantic feature correspondences between target meshes. This paper proposes a novel approach for parameterizing a mesh into another one directly. The main idea of our method is to combine a competitive learning and a least-square mesh techniques. It is enough to give some semantic feature correspondences between target meshes, even if they are in different shapes or in different poses.

In general, in competitive learning the requirement for the initial number of prototypes is a difficult task, as we do not usually know the number of clusters in the input data a priori. The behavior and performance of the competitive algorithms are very sensitive to the initial locations and number of the prototypes. In this paper after investigating several important competitive learning paradigms, we present compensation techniques for overcoming the problems in competitive learning. Our experimental results show that competition with compensation can improve the performance of the learning algorithm.

In the handwritten character recognition, it is very important to extract essential structure of character patterns. Requirements for skeletonization can be summarized as follows: (a) Insensitive to irregular edge lines. (b) Nonstructure patterns are not extracted. (c) Insensitive to nonuniform line width. (d) Line information is held. In this paper, a weighted competitive learning method is proposed in order to achieve the above requirements. Regarding (a) and (b), unnecessary pattern information is removed by representing some region of the pattern using a single representative point (RP). In order to optimize the RPs, the competitive learning is employed. For the requirement (c), the region, covered by a RP, is adjusted according to the line width. The condition (d) is satisfied by connecting the RPs along the line and also through the border of the regions. Simulation results, obtained using so many kinds of distorted patterns, including digits, alphabet and Japanese Kanji, demonstrate the proposed method can extract essential skeleton structure despite of several distortions.

This paper presents a novel pipelined architecture for fast competitive learning (CL). It is used as a hardware accelerator in a system on programmable chip (SOPC) for reducing the computational time. In the architecture, a novel codeword swapping scheme is adopted so that both neuron competition processes for different training vectors can be operated concurrently. The neuron updating process is based on a hardware divider with simple table lookup operations. The divider performs finite precision calculation for area cost reduction at the expense of slight degradation in training performance. Experimental results show that the CPU time is lower than that of other hardware or software implementations running the CL training program with or without the support of custom hardware.

An unsupervised competitive learning algorithm based on the classical -means clustering algorithm is proposed. The proposed learning algorithm called the centroid neural network (CNN) estimates centroids of the related cluster groups in training date. This paper also explains algorithmic relationships among the CNN and some of the conventional unsupervised competitive learning algorithms including Kohonen's self-organizing map (SOM) and Kosko's differential competitive learning (DCL) algorithm. The CNN algorithm requires neither a predetermined schedule for learning coefficient nor a total number of iterations for clustering. The simulation results on clustering problems and image compression problems show that CNN converges much faster than conventional algorithms with compatible clustering quality while other algorithms may give unstable results depending on the initial values of the learning coefficient and the total number of iterations.

Cryogenic electron tomography (cryo-ET) visualizes the 3D spatial distribution of macromolecules at nanometer resolution inside native cells. However, automated identification of macromolecules inside cellular tomograms is challenged by noise and reconstruction artifacts, as well as the presence of many molecular species in the crowded volumes. Here, we present DeepFinder, a computational procedure that uses artificial neural networks to simultaneously localize multiple classes of macromolecules. Once trained, the inference stage of DeepFinder is faster than template matching and performs better than other competitive deep learning methods at identifying macromolecules of various sizes in both synthetic and experimental datasets. On cellular cryo-ET data, DeepFinder localized membrane-bound and cytosolic ribosomes (roughly 3.2 MDa), ribulose 1,5-bisphosphate carboxylase-oxygenase (roughly 560 kDa soluble complex) and photosystem II (roughly 550 kDa membrane complex) with an accuracy comparable to expert-supervised ground truth annotations. DeepFinder is therefore a promising algorithm for the semiautomated analysis of a wide range of molecular targets in cellular tomograms.

This article explores the application of artificial intelligence (AI) technology in Go level examinations and its impact on Go education and evaluation systems. With the advancement of science and technology, especially the rapid development of AI technology, Go, as an ancient strategic game, has experienced revolutionary changes in its competitive scenes and learning methods. After AlphaGo defeated world Go champion Lee Sedol in 2016, the application of AI in the field of Go received widespread attention. AI has not only demonstrated capabilities beyond humans in high-level Go games, but has also played an increasingly important role in Go teaching and training. The global spread of the COVID-19 epidemic has forced Go level examinations to find new ways to adapt. The Go online intelligent test came into being, providing a new, efficient and objective assessment method. Although the introduction of online intelligent examinations has brought convenience, it has also raised issues in organizational structure adjustment, revenue distribution, and examination fairness. Challenges faced by the Go online intelligence examination in terms of fairness, including lack of face-to-face supervision, possible cheating, etc., pose new challenges to the fairness and authority of the Go level examination. This article compares the differences between intelligent Go exams and traditional exams, including assessment format, exam content and passing rate. The online intelligent examination has added test questions and raised the requirements for candidates’ computing power. It also shows a higher passing rate than traditional methods. However, the authenticity and fairness of online intelligent examinations have been challenged, including the adaptability of AI game difficulty and the lack of examination supervision, leading to the emergence of cheating. The positive impacts of implementing online smart exams include improving the convenience and efficiency of exams, simplifying event organization, and promoting the development of Go players’ skills. But at the same time, this new form has had an impact on the original Go organizational structure and benefit distribution, causing concerns and adjustments from local associations. In order to solve these problems, this article puts forward a series of suggestions, including setting up a supervision mechanism for intelligent examinations, optimizing the assessment system, and adjusting the revenue distribution ratio to ensure the long-term sustainable development of online intelligent examinations. In summary, the use of artificial intelligence in Go level examinations not only reflects the persistence of the interaction between technology and traditional cultural activities, but also poses ongoing challenges to the fairness and validity of education and assessment methods. By deeply understanding these impacts and taking corresponding countermeasures, we can find a balance between innovation and tradition, and provide support for the development of Go education and evaluation systems.

Multi-label learning aims at assigning a set of appropriate labels to multi-label samples. Although it has been successfully applied in various domains in recent years, most multi-label learning methods require sufficient labeled training samples, because of the large number of possible label sets. Co-training, as an important branch of semi-supervised learning, can leverage unlabeled samples, along with scarce labeled ones, and can potentially help with the large labeled data requirement. However, it is a difficult challenge to combine multi-label learning with co-training. Two distinct issues are associated with the challenge: (i) how to solve the widely-witnessed class-imbalance problem in multi-label learning; and (ii) how to select samples with confidence, and communicate their predicted labels among classifiers for model refinement. To address these issues, we introduce an approach called Multi-Label Co-Training (MLCT). MLCT leverages information concerning the co-occurrence of pairwise labels to address the class-imbalance challenge; it introduces a predictive reliability measure to select samples, and applies label-wise filtering to confidently communicate labels of selected samples among co-training classifiers. MLCT performs favorably against related competitive multi-label learning methods on benchmark datasets and it is also robust to the input parameters.

Inspired by the human learning principle that learning easier concepts first and then gradually paying more attention to harder ones, curriculum learning uses the nonuniform sampling of mini-batches according to the order of examples’ difficulty. Just as a teacher adjusts the curriculum according to the learning progress of each student, a proper curriculum should be adapted to the current state of the model. Therefore, in contrast to recent works using a fixed curriculum, we devise a new curriculum learning method, Adaptive Curriculum Learning (Adaptive CL), adapting the difficulty of examples to the current state of the model. Specifically, we make use of the loss of the current model to adjust the difficulty score while retaining previous useful learned knowledge by KL divergence. Moreover, under a non-linear model and binary classification, we theoretically prove that the expected convergence rate of curriculum learning monotonically decreases with respect to the loss of a point regarding the optimal hypothesis, and monotonically increases with respect to the loss of a point regarding the current hypothesis. The analyses indicate that Adaptive CL could improve the convergence properties during the early stages of learning. Extensive experimental results demonstrate the superiority of the proposed approach over existing competitive curriculum learning methods.

Segmentation of images into textural homogeneous regions is a fundamental problem in an image understanding system. Most region-oriented segmentation approaches suffer from the problem of different thresholds selecting for different images. In this paper an adaptive image segmentation based on vector quantization is presented. It automatically segments images without preset thresholds. The approach contains a feature extraction module and a two-layer hierarchical clustering module, a vector quantizer (VQ) implemented by a competitive learning neural network in the first layer. A near-optimal competitive learning algorithm (NOLA) is employed to train the vector quantizer. NOLA combines the advantages of both Kohonen self- organizing feature map (KSFM) and K-means clustering algorithm. After the VQ is trained, the weights of the network and the number of input vectors clustered by each neuron form a 3- D topological feature map with separable hills aggregated by similar vectors. This overcomes the inability to visualize the geometric properties of data in a high-dimensional space for most other clustering algorithms. The second clustering algorithm operates in the feature map instead of the input set itself. Since the number of units in the feature map is much less than the number of feature vectors in the feature set, it is easy to check all peaks and find the `correct' number of clusters, also a key problem in current clustering techniques. In the experiments, we compare our algorithm with K-means clustering method on a variety of images. The results show that our algorithm achieves better performance.

The world of education has undergone a major shift as a result of advances in information and communication technology. This study compares conventional and digital learning methods for distance learning (PJJ) across the board, with the aim of finding out what is effective, what is not effective, and how the two methods affect student learning outcomes and experiences. Conventional learning, which relies on direct interaction between teachers and students, emphasizes elements such as giving feedback, easy to do, structured, and routine. It has the advantage of being able to facilitate direct interpersonal communication, provide intensive supervision and guidance, and create a competitive learning environment. However, this method has some limitations, especially in terms of time and space flexibility. It is also not able to meet the learning needs of individuals in large groups. On the other hand, digital learning offers greater flexibility, access to a wide range of educational resources, and increased interactivity through the use of online collaboration tools, e-learning platforms, and video conferencing. The use of technology allows learning content to be presented in a systematic, structured and engaging manner, increasing student motivation and participation. In addition, it allows materials to be tailored to each student's learning style and conditions, and supports collaborative and independent learning. Challenges include the possibility of misuse of technology, limited access in some areas, lack of direct social interaction, and reliance on digital infrastructure and capabilities. The research results indicate that digital learning generally enhances student learning outcomes more significantly compared to conventional learning, especially in terms of distance learning (PJJ). Digital strategies such as structured instructional design, integration of practical experiences, motivational support, and time management have been shown to improve student interaction, understanding, and independence. However, digital and conventional learning methods can complement each other to achieve optimal effectiveness. The application of these strategies not only boosts students' academic performance but also equips them with the essential skills to tackle future challenges.