Disentangled representation learning aims to decompose images into meaningful independent factors of variation. However, measuring the extent of disentanglement remains a challenge. Available measures, such as β-VAE, MIG, SAP score or Explicitness score rely on classifier assumptions, sampling schemes, or mutual information estimators, introducing biases and dependencies to the model. In this paper, we present a novel correlation-based measure, CorDis, which mitigates these dependencies while preserving robust and interpretable insights into disentanglement. We systematically compare CorDis with existing measures. Experimental results demonstrate that CorDis provides a more principled and assumption-light approach to measuring the amount of disentanglement, contributing to the development of universal and consistent benchmarks.

Abstract Although Convolutional Neural Networks (CNN) outperform the classical models in a wide range of Machine Vision applications, their restricted interpretability and their lack of comprehensibility in reasoning, generate many problems such as security, reliability, and safety. Consequently, there is a growing need for research to improve explainability and address their limitations. In this paper, we propose a concept-based method, called Concept-Aware Explainability (CAE) to provide a verbal explanation for the predictions of pre-trained CNN models. A new measure, called detection score mean, is introduced to quantify the relationship between the filters of the model and a set of pre-defined concepts. Based on the detection score mean values, we define sorted lists of Concept-Aware Filters (CAF) and Filter-Activating Concepts (FAC). These lists are used to generate explainability reports, where we can explain, analyze, and compare models in terms of the concepts embedded in the image. The proposed explainability method is compared to the state-of-the-art methods to explain Resnet18 and VGG16 models, pre-trained on ImageNet and Places365-Standard datasets. Two popular metrics, namely, the number of unique detectors and the number of detecting filters, are used to make a quantitative comparison. Superior performances are observed for the suggested CAE, when compared to Network Dissection (NetDis) (Bau et al., in: Proceedings of the IEEE conference on computer vision and pattern recognition (CVPR), 2017), Net2Vec (Fong and Vedaldi, in: Paper presented at IEEE conference on computer vision and pattern recognition (CVPR), 2018), and CLIP-Dissect (CLIP-Dis) (Oikarinen and Weng, in: The 11th international conference on learning representations (ICLR), 2023) methods.

In this work, we propose two novel methods to align the feature representation of fMRI data to improve the transfer learning performance on brain decoding problem under a limited data regime. Transfer learning on brain decoding is most viable when the source and target datasets share the same feature space. In such a condition, the feature alignment method should prioritize dominant variance directions of data. Following this requirement, in our first method, we modify the standard formulation of maximum variance generalized canonical correlation analysis (maxvar-GCCA) to work on linear projectors, scaled by eigenvalues. Our method suppresses the low variance dimensions of the feature space, which, otherwise, can spoil the feature alignment process. We estimate a series of linear transformations, using principal component analysis to align hierarchically organized data at session, subject, and dataset levels. In the second method, we incorporate label information of fMRI recordings for better transferable features in a limited data regime. We extend our first method to utilize correlation between labels and data representations. Our solution imposes label dependence by maximizing the Hilbert Schmidt Independence Criterion as an intermediate step in our hierarchical feature alignment procedure. Both methods are applicable to heterogeneously labeled time samples, which do not impose the same label at each time instant of different sessions or require an equal number of time points across the sessions. We compare our methods with the state-of- the-art solution in transfer learning methods for neural decoding that follows the method of maxvar-GCCA. We evaluate our transfer learning methods using four non-intersecting stop-signal paradigm datasets. We observe that the suggested Hierarchical Feature Alignment method and its label-guided version improve the state-of-the-art transfer learning method for brain decoding consistently on all transfer learning cases.

One way of designing a robust machine learning algorithm is to generate authentic adversarial images that can trick the algorithms as much as possible. In this study, we propose a new method to generate adversarial images which are very similar to true images, yet, these images are discriminated from the original ones and are assigned into another category by the model. The proposed method is based on a popular concept of experimental psychology, called, Just Noticeable Difference. We define Just Noticeable Difference for a machine learning model and generate the least perceptible difference for adversarial images which can trick a model. The suggested model iteratively distorts a true image by gradient descent method until the machine learning algorithm outputs a false label. Deep Neural Networks are trained for object detection and classification tasks. The cost function includes regularization terms to generate just noticeably different adversarial images which can be detected by the model. The adversarial images generated in this study look more natural compared to the output of the state of the art adversarial image generators.

Representing brain activities by networks is very crucial to understand various cognitive states. This study proposes a novel method to estimate static and dynamic brain networks using Kulback-Leibler divergence. The suggested brain networks are based on the probability distributions of voxel intensity values measured by functional Magnetic Resonance Images (fMRI) recorded while the subjects perform a predefined cognitive task, called complex problem solving. We investigate the validity of the estimated brain networks by modeling and analyzing the different phases of complex problem solving process of human brain, namely planning and execution phases. The suggested computational network model is tested by a classification schema using Support Vector Machines. We observe that the network models can successfully discriminate the planning and execution phases of complex problem solving process with more than 90% accuracy, when the estimated dynamic networks, extracted from the fMRI data, are classified by Support Vector Machines.

In this paper, we aim to measure the information content of brain anatomic regions using the functional magnetic resonance images (fMRI) recorded during a complex problem solving (CPS) task. We, also, analyze the brain regions, activated in different phases of the problem solving process. Previous studies have widely used machine learning approaches to examine the active anatomic regions for cognitive states of human subjects based on their fMRI data. This study proposes an information theoretic method for analyzing the activity in anatomic regions. Briefly, we define and estimate two types of Shannon entropy, namely, static and dynamic entropy, to understand how complex problem solving processes lead to changes in information content of anatomic regions. We investigate the relationship between the problem-solving task phases and the Shannon entropy measures suggested in this study, for the underlying brain activity during a Tower of London (TOL) problem solving process. We observe that the dynamic entropy fluctuations in brain regions during the CPS task provides a measure for the information content of the main phases of complex problem solving, namely planning and execution. We, also, observe that static entropy measures of anatomic regions are consistent with the experimental findings of neuroscience. The preliminary results show strong promise in using the suggested static and dynamic entropy as a measure for characterizing the brain states related to the problem solving process. This capability would be useful in revealing the hidden cognitive states of subjects performing a specific cognitive task.

We present a non-Euclidean vector product for artificial neural networks. The vector product operator does not require any multiplications while providing correlation information between two vectors. Ordinary neurons require inner product of two vectors. We propose a class of neural networks with the universal approximation property over the space of Lebesgue integrable functions based on the proposed non-Euclidean vector product. In this new network, the “product” of two real numbers is defined as the sum of their absolute values, with the sign determined by the sign of the product of the numbers. This “product” is used to construct a vector product in RN. The vector product induces the l1norm. The additive neural network successfully solves the XOR problem. Experiments on MNIST and CIFAR datasets show that the classification performance of the proposed additive neural network is comparable to the corresponding multi-layer perceptron and convolutional neural networks.

In this study, we propose a correlation based variational change detection (CVCD) method for elevation models. In essence, CVCD aims to produce smooth change maps while preserving the details of the terrain by minimizing a variational cost function. Proposed cost function is constructed with a novel data fidelity term using normalized correlation coefficient and l1 -norm total variation regularization term. An effective algorithm is suggested to minimize the cost function using simple approximations in an iterative method. Quantitative experiments on synthetic noisy data show that CVCD can provide a detection rate of 95% while staying in the low false alarm regime, i.e. less than 10-2. Also, qualitative experiments on real-world data show the success of the CVCD for the changes with different characteristics.

The main goal of this study is to extract a set of brain networks in multiple time-resolutions to analyze the connectivity patterns among the anatomic regions for a given cognitive task. We suggest a deep architecture which learns the natural groupings of the connectivity patterns of human brain in multiple time-resolutions. The suggested architecture is tested on task data set of Human Connectome Project (HCP) where we extract multi-resolution networks, each of which corresponds to a cognitive task. At the first level of this architecture, we decompose the fMRI signal into multiple sub-bands using wavelet decompositions. At the second level, for each sub-band, we estimate a brain network extracted from short time windows of the fMRI signal. At the third level, we feed the adjacency matrices of each mesh network at each time-resolution into an unsupervised deep learning algorithm, namely, a Stacked De- noising Auto-Encoder (SDAE). The outputs of the SDAE provide a compact connectivity representation for each time window at each sub-band of the fMRI signal. We concatenate the learned representations of all sub-bands at each window and cluster them by a hierarchical algorithm to find the natural groupings among the windows. We observe that each cluster represents a cognitive task with a performance of 93% Rand Index and 71% Adjusted Rand Index. We visualize the mean values and the precisions of the networks at each component of the cluster mixture. The mean brain networks at cluster centers show the variations among cognitive tasks and the precision of each cluster shows the within cluster variability of networks, across the subjects.

In this paper, we propose a novel finetuning algorithm for the recently introduced multiway, multilingual neural machine translate that enables zero-resource machine translation.When used together with novel manyto-one translation strategies, we empirically show that this finetuning algorithm allows the multi-way, multilingual model to translate a zero-resource language pair (1) as well as a single-pair neural translation model trained with up to 1M direct parallel sentences of the same language pair and (2) better than pivotbased translation strategy, while keeping only one additional copy of attention-related parameters.

Attack detection problems in the smart grid are posed as statistical learning problems for different attack scenarios in which the measurements are observed in batch or online settings. In this approach, machine learning algorithms are used to classify measurements as being either secure or attacked. An attack detection framework is provided to exploit any available prior knowledge about the system and surmount constraints arising from the sparse structure of the problem in the proposed approach. Well-known batch and online learning algorithms (supervised and semisupervised) are employed with decision- and feature-level fusion to model the attack detection problem. The relationships between statistical and geometric properties of attack vectors employed in the attack scenarios and learning algorithms are analyzed to detect unobservable attacks using statistical learning methods. The proposed algorithms are examined on various IEEE test systems. Experimental analyses show that machine learning algorithms can detect attacks with performances higher than attack detection algorithms that employ state vector estimation methods in the proposed attack detection framework.

Learning low dimensional embedding spaces (manifolds) for efficient feature representation is crucial for complex and high dimensional input spaces. Functional magnetic resonance imaging (fMRI) produces high dimensional input data and with a less then ideal number of labeled samples for a classification task. In this study, we explore deep learning methods for fMRI classification tasks in order to reduce dimensions of feature space, along with improving classification performance for brain decoding. We employ sparse autoencoders for unsupervised feature learning, leveraging unlabeled fMRI data to learn efficient, non-linear representations as the building blocks of a deep learning architecture by stacking them. Proposed method is tested on a memory encoding/retrieval experiment with ten classes. The results support the efficiency compared to the baseline multi-voxel pattern analysis techniques.

The performance of object recognition and classification on remote sensing imagery is highly dependent on the quality of extracted features, amount of labelled data and the priors defined for contextual models. In this study, we examine the representation learning opportunities for remote sensing. First we attacked localization of contextual cues for complex object detection using disentangling factors learnt from a small amount of labelled data. The complex object, which consists of several sub-parts is further represented under the Conditional Markov Random Fields framework. As a second task, end-to-end target detection using convolutional sparse auto-encoders (CSA) using large amount of unlabelled data is analysed. Proposed methodologies are tested on complex airfield detection problem using Conditional Random Fields and recognition of dispersal areas, park areas, taxi routes, airplanes using CSA. The method is also tested on the detection of the dry docks in harbours. Performance of the proposed method is compared with standard feature engineering methods and found competitive with currently used rule-based and supervised methods.

In this study, a fast and efficient consensus segmentation method is proposed which fuses a set of baseline segmentation maps under an unsupervised Markov Random Fields (MRF) framework. The degree of consensus among the segmentation maps are estimated as the relative frequency of co occurrences among the adjacent segments. Then, these relative frequencies are used to construct the energy function of an unsupervised MRF model. It is well-known that MRF framework is commonly used for formulating the spatial relationships among the super-pixels, under the Potts model. In this study, the Potts model is reorganized to represent the degree of consensus among the spatially adjacent segments (super-pixels). The proposed segmentation fusion method, called, Boosted-MRF, is tested in various experimental setups, and its performance is compared to the state of the art segmentation methods and satisfactory results are obtained.

An information theoretic approach is proposed to estimate the degree of connectivity for each voxel with its neighboring voxels. The neighborhood system is defined by spatial and functional connectivity metrics. Then, a local mesh of variable size is formed around each voxel using spatial or functional neighborhood. The mesh arc weights, called Mesh Arc Descriptors (MAD), are estimated by a linear regression model fitted to the voxel intensity values of the functional Magnetic Resonance Images (fMRI). Finally, the error term of the linear regression equation is used to estimate the mesh size for a voxel by optimizing Akaike's information Criterion, Bayesian Information Criterion and Rissanen's Minimum Description Length. fMRI measurements are obtained during a memory encoding and retrieval experiment performed on a subject who is exposed to the stimuli from 10 semantic categories. For each sample, a k-NN classifier is trained using the Mesh Arc Descriptors (MAD) having the variable mesh sizes. The classification performances reflect that the suggested variable-size Mesh Arc Descriptors represents the mental states better than the classical multi-voxel pattern representation. Moreover, we observe that the degree of connectivities in the brain greatly varies for each voxel.

Segment based classification is one of the popular approaches for object detection, where the performance of the classification task is sensitive to the accuracy of the output of the initial segmentation. Most of these studies includes generic segmentation methods and it is assumed that the segmentation output is compatible with the subsequent classification method. However, depending on the problem domain the properties of the regions such as size, shape etc. which are suitable for classification may vary. In this study, we propose a domain specific segmentation method for building detection. The contribution of the domain specific segmentation is empirically analyzed. For this purpose, first the decision fusion method is employed for building detection on the outputs of the state of the art segmentation methods, then it is employed on the output of the domain specific segmentation method and the classification performances for each method are compared. The advantage of domain specific segmentation is observed quantitatively and satisfactory results are obtained.

In this study, an information theoretic approach is proposed to model brain connectivity during a cognitive processing task, measured by functional Magnetic Resonance Imaging (fMRI). For this purpose, a local mesh of varying size is formed around each voxel. The arc weights of each mesh are estimated using a linear regression model by minimizing the squared error. Then, the optimal mesh size for each sample, that represents the information distribution in the brain, is estimated by minimizing various information criteria which employ the mean square error of linear regression model. The estimated mesh size shows the degree of locality or degree of connectivity of the voxels for the underlying cognitive process. The samples are generated during an fMRI experiment employing item recognition (IR) and judgment of recency (JOR) tasks. For each sample, estimated arc weights of the local mesh with optimal size are used to classify whether it belongs to IR or JOR tasks. Results indicate that the suggested connectivity model with optimal mesh size for each sample represent the information distribution in the brain better than the state-of-the art methods.

Machine learning algorithms have been widely used as reliable methods for modeling and classifying cognitive processes using functional Magnetic Resonance Imaging (fMRI) data. In this study, we aim to classify fMRI measurements recorded during an object recognition experiment. Previous studies focus on Multi Voxel Pattern Analysis (MVPA) which feeds a set of active voxels in a concatenated vector form to a machine learning algorithm to train and classify the cognitive processes. In most of the MVPA methods, after an image preprocessing step, the voxel intensity values are fed to a classifier to train and recognize the underlying cognitive process. Sometimes, the fMRI data is further processed for de-noising or feature selection where techniques, such as Generalized Linear Model (GLM), Independent Component Analysis (ICA) or Principal Component Analysis are employed. Although these techniques are proved to be useful in MVPA, they do not model the spatial connectivity among the voxels. In this study, we attempt to represent the local relations among the voxel intensity values by forming a mesh network around each voxel to model the relationship of a voxel and its surroundings. The degree of connectivity of a voxel to its surroundings is represented by the arc weights of each mesh. The arc weights, which are estimated by a linear regression model, are fed to a classifier to discriminate the brain states during an object recognition task. This approach, called Mesh Learning, provides a powerful tool to analyze various cognitive states using fMRI data. Compared to traditional studies which focus either merely on multi-voxel pattern vectors or their reduced-dimension versions, the suggested Mesh Learning provides a better representation of object recognition task. Various machine learning algorithms are tested to compare the suggested Mesh Learning to the state-of-the art MVPA techniques. The performance of the Mesh Learning is shown to be higher than that of the available MVPA techniques.

A new segmentation fusion method is proposed that ensembles the output of several segmentation algorithms applied on a remotely sensed image. The candidate segmentation sets are processed to achieve a consensus segmentation using a stochastic optimization algorithm based on the Filtered Stochastic BOEM (Best One Element Move) method. For this purpose, Filtered Stochastic BOEM is reformulated as a segmentation fusion problem by designing a new distance learning approach. The proposed algorithm also embeds the computation of the optimum number of clusters into the segmentation fusion problem.

New methods that exploit sparse structures arising in smart grid networks are proposed for the state estimation problem when data injection attacks are present. First, construction strategies for unobservable sparse data injection attacks on power grids are proposed for an attacker with access to all network information and nodes. Specifically, novel formulations for the optimization problem that provide a flexible design of the trade-off between performance and false alarm are proposed. In addition, the centralized case is extended to a distributed framework for both the estimation and attack problems. Different distributed scenarios are proposed depending on assumptions that lead to the spreading of the resources, network nodes and players. Consequently, for each of the presented frameworks a corresponding optimization problem is introduced jointly with an algorithm to solve it. The validity of the presented procedures in real settings is studied through extensive simulations in the IEEE test systems.

This paper proposes a novel approach to building detection problem in satellite images. The proposed method employs a two layer hierarchical classification mechanism for ensemble learning. After an initial segmentation, each segment is classified by N different classifiers using different features at the first layer. The class membership values of the segments, which are obtained from different base layer classifiers, are ensembled to form a new fusion space, which forms a linearly separable simplex. Then, this simplex is partitioned by a linear classifier at the meta layer. The paper presents the performance results of the proposed model and comparisons with the state of the art classifiers.

In this study, we introduce a new image classification technique using decision fusion. The proposed technique, called Meta-Fuzzified Yield Value (Meta-FYV), is based on two-layer Stacked Generalization (SG) architecture. At the base-layer, the system, receives a set of feature vectors of various dimensions and dynamical ranges and outputs hypotheses through fuzzy transformations. Then, the hypotheses created by the base layer transformations are concatenated for building a regression equation at meta-layer. Experimental evidence indicates that the Meta-FYV is superior compared to one of the most successful Fuzzy SG methods, introduced by Akbas.

Automatic image annotation systems available in the literature concatenate color, texture and/or shape features in a single feature vector to learn a set of high level semantic categories using a single learning machine. This approach is quite naive to map the visual features to high level semantic information concerning the categories. Concatenation of many features with different visual properties and wide dynamical ranges may result in curse of dimensionality and redundancy problems. Additionally, it usually requires normalization which may cause an undesirable distortion in the feature space. An elegant way of reducing the effects of these problems is to design a dedicated feature space for each image category, depending on its content, and learn a range of visual properties of the whole image from a variety of feature sets. For this purpose, a two-layer ensemble learning system, called Supervised Annotation by Descriptor Ensemble (SADE), is proposed. SADE, initially, extracts a variety of low-level visual descriptors from the image. Each descriptor is, then, fed to a separate learning machine in the first layer. Finally, the meta-layer classifier is trained on the output of the first layer classifiers and the images are annotated by using the decision of the meta-layer classifier. This approach not only avoids normalization, but also reduces the effects of dimensional curse and redundancy. The proposed system outperforms a state-of-the-art automatic image annotation system, in an equivalent experimental setup.

In this study, we propose a hierarchical classification system, which emulates the eye-brain channel in two hierarchical layers. In the first layer, a set of classifiers are trained by using low level, low dimensional features. In the second layer, the recognition results of the first layer are fed to the Fuzzy ARTMAP (FAM) classifier which implements the Adaptive Resonance Theory. Experiments indicate that the hierarchical approach proposed in this paper, increases the classification performances compared to the available methods.

This study deals with designing a flexible feature space for Content Based Image retrieval Systems (CBIR). For this purpose, initially, a large variety of features are extracted from the regions of the pre-segmented images. Then, the feature set of each object class is learned using the Fuzzy Art Map Architecture, by identifying the weights of each feature for each object class.

We suggest a method to adapt an image retrieval system into a configurable one. Basically, the original feature space of a content-based retrieval system is nonlinearly transformed into a new space, where the distance between the feature vectors is adjusted by learning. The transformation is realized by an artificial neural network architecture. A cost function is defined for learning and optimized by the simulated annealing method. Experiments are done on the texture image retrieval system, which use Gabor filter features. The results indicate that configured image retrieval system is significantly better than the original system.

A new analytic scheme, which uses a sequence of image segmentation and recognition algorithms, is proposed for the off-line cursive handwriting recognition problem. First, some global parameters, such as slant angle, baselines, stroke width and height, are estimated. Second, a segmentation method finds character segmentation paths by combining gray-scale and binary information. Third, a hidden Markov model (HMM) is employed for shape recognition to label and rank the character candidates. For this purpose, a string of codes is extracted from each segment to represent the character candidates. The estimation of feature space parameters is embedded in the HMM training stage together with the estimation of the HMM model parameters. Finally, information from a lexicon and from the HMM ranks is combined in a graph optimization problem for word-level recognition. This method corrects most of the errors produced by the segmentation and HMM ranking stages by maximizing an information measure in an efficient graph search algorithm. The experiments indicate higher recognition rates compared to the available methods reported in the literature.

A new texture descriptor, namely, statistical analysis of structural information (SASI) is introduced as a representation of texture. SASI is based on the statistics of clique autocorrelation functions calculated over a set of directional moving windows. SASI defines a set of windows to extract and measure various structural properties of texture by using a spatial multiresolution method. Although it works in the spatial domain, it measures the spectral information of a given texture. Experimental results, performed on the digitized Brodatz Album, indicate that SASI is very successful in identifying "similar" textures.

Character recognition (CR) has been extensively studied in the last half century and has progressed to a level that is sufficient to produce technology-driven applications. Now, rapidly growing computational power is enabling the implementation of the present CR methodologies and is creating an increasing demand in many emerging application domains which require more advanced methodologies. This paper serves as a guide and update for readers working in the CR area. First, the historical evolution of CR systems is presented. Then, the available CR techniques, with their superiorities and weaknesses, are reviewed. Finally, the current status of CR is discussed and directions for future research are suggested. Special attention is given to off-line handwriting recognition, since this area requires more research in order to reach the ultimate goal of machine simulation of human reading.

Currently, image retrieval systems are based on low-level features of color, texture and shape, not on the semantic descriptions that are common to humans, such as objects, people, and place. In order to narrow down the gap between the low level and semantic level, object-based content analysis, which segments the semantically meaningful objects of images, is an essential step. In this study, we propose a learning process in order to perform effective automatic off-line analysis on a multi-level segmented image stack. Meaningful objects are extracted given certain user search patterns and interest profiles. Color and/or shape information of the objects is stored in the hierarchical content representations of the images. This information is utilized by a hierarchical matching scheme to improve the retrieval speed in the subsequent searches.

Given the shape information of an object, can we find visually meaningful "n" objects in an image database, which is ranked from the most similar to the n/sup th/ similar one? The answer to this question depends on the complexity of the images in the database and the complexity of the objects in the query. This study presents a robust shape descriptor, which compares a given object to the objects in an image database and identifies "n" shapes, ranked from the most similar to the least similar one, in the database. The intended shape descriptor is based on the circular hidden Markov model (HMM) proposed by the authors (1999) of this study. The circular HMM is both ergodic and temporal. It is insensitive to size changes. Since it has no starting and terminating state, it is insensitive to the starting point of the shape boundary. The experiments, performed on 100 test shapes, indicate excellent result.

We introduce a new set of one-dimensional discrete, constant length features to represent two dimensional shape information for HMM (hidden Markov model), based handwritten optical character recognition problem. The proposed feature set embeds the two dimensional information into a sequence of one-dimensional codes, selected from a code book. It provides a consistent normalization among distinct classes of shapes, which is very convenient for HMM based shape recognition schemes. The new feature set is used in a handwritten optical character recognition scheme, where a sequence of segmentation and recognition stages is employed. The normalization parameters, which maximize the recognition rate, are dynamically estimated in the training stage of the HMM. The proposed character recognition system is tested on both a locally generated cursively handwritten data and isolated number digits of NIST database. The experimental results indicate high recognition rates.

In this study a robust color quantization method, which is based on a one-dimensional dynamic thresholding method, is introduced. The proposed method extracts a set of one-dimensional color intervals, each of which is ordered with respect to the distance to the reference color of that interval. The color intervals are then used to form the rows of the color similarity matrix for a given image. The selection of color palate is accomplished on the color similarity matrix by minimizing the total square error with respect to a threshold variable, which dynamically defines the color similarity for a given image. The experimental results indicate that the proposed method yields smaller quantization error and better visual appearance compared to the Heckbert's algorithm. It is faster than the existing color quantization methods.

A scheme is proposed for off-line handwritten connected digit recognition, which uses a sequence of segmentation and recognition algorithms. First, the connected digits are segmented by employing both the gray scale and binary information. Then, a new set of features is extracted from the segments. The parameters of the feature set are adjusted during the training stage of the hidden Markov model (HMM) where the potential digits are recognized. Finally, in order to confirm the preliminary segmentation and recognition results, a recognition based segmentation method is presented.

We introduce a scheme for an off-line handwritten connected digit string recognition problem, which uses a sequence of segmentation and recognition algorithms. The proposed system assumes no constraint in writing style, size or variations. First, a segmentation method, which combines the gray scale and binary information, is proposed to find the nonlinear character segmentation paths. Each segment is then, recognized by a hidden Markov model. Finally, in order to confirm the segmentation paths and recognition results, a recognition based segmentation method is presented. The proposed scheme is tested on 4000 handwritten connected digits, collected from 16 different persons. The experiments yield 97.2% recognition rate.

This paper describes a new framework, called repulsive attractive (RA) network for baseline extraction on document images. The RA network is a self organizing feature detector which interacts with the document text image through the attractive and repulsive forces defined among the network components and the document image. Experimental results indicate that the network can successfully extract the baselines under heavy noise and with overlaps between the ascending and descending portions of the characters of adjacent lines. The proposed method is also applicable to a wide range of image processing applications, such as curve fitting, segmentation and thinning.

A segmentation-free approach for off-line optical character recognition is presented. The proposed method performs the recognition by extracting the characters from the whole word, avoiding the segmentation process. A control point set which includes position and attribute vectors is selected for the features. In the training mode, each sample character is mapped to a set of control points and is stored in an archive which belongs to an alphabet. In the recognition mode, the control points of the input image are first extracted. Then, each control point is matched to the control points in the alphabet according to its attributes. During the matching process, a probability matrix is constructed which holds some matching measures (probabilities) for identifying the characters. Experimental results indicate that the proposed method is very robust in extracting the characters from a cursive script.