Latika Singh

A hybrid feature selection approach based on improved PSO and filter approaches for image steganalysis

This paper proposes a novel feature selection approach to improve the classification accuracy and reduce the computational complexity in image steganalysis. It is a hybrid filter-wrapper approach based on improved Particle Swarm Optimization (PSO). It consists of two phases: the first phase is composed of two filter techniques namely t test and multiple-regression which selects the features based on their ability to discriminate images as stego or cover. The second phase further reduces the number of features by working on the significant features selected during the first phase using an improved PSO. This approach overcomes the disadvantages of global best PSO by integrating it with local best PSO and dynamically changing the population size (Hope/Rehope). The proposed approach is tested on two sets of features extracted from spatial domain (SPAM-Subtractive Adjacency Matrix) and transform domain (CCPEV-Cartesian Calibrated features extracted by Pevný) for four embedding algorithms nsF5, Outguess, Perturbed Quantization and Steghide using SVM (Support Vector Machine) classifier. Experimental results demonstrate that this approach significantly improves the classification accuracy and drastically reduces dimensionality as compared to results produced by some well-known feature selection algorithms.

An Improved Discrete Firefly and t-Test based Algorithm for Blind Image Steganalysis

Feature Selection is a preprocessing technique with great significance in data mining applications that aims at reducing computational complexity and increase predictive capability of a learning system. This paper presents a new hybrid feature selection algorithm based on Discrete Firefly optimization technique with dynamic alpha and gamma parameters and t-test filter technique to improve detectability of hidden message for Blind Image Steganalysis. The experiments are conducted on important dataset of feature vectors extracted from frequency domain, Discrete Cosine Transformation and Discrete Wavelet Transformation domain of cover and stego images. The results from popular JPEG steganography algorithms nsF5, Outguess, PQ and JP Hide and Seek show that proposed method is able to identify sensitive features and reduce the feature set by 67% in DCT domain and 37% in DWT domain. The experiment analysis shows that these algorithms are most sensitive to Markov features from DCT domain and variance statistical moment from DWT domain.

Developmental pattern analysis and age prediction by extracting speech features and applying various classification techniques

In speech development research, its important to know how speech acoustic features vary as a function of age and the age when the variability and magnitude of acoustic features start to exhibit adult-like patterns. During the first few years of life, a childs speech changes from the cries and babbles of an infant to adult-like words and phrases of a young child. A number of acoustic studies observed that, adults speech compared to childrens speech, exhibits lower pitch and formant frequencies, shorter segmental durations, and lesser temporal and spectral variability. In this research we extracted acoustic, spectral and temporal features of a speech signal and then classify these features to predict the age of the subjects using different classification techniques. The feature vector comprised of fundamental frequency, formants, lpc coefficients and segmental duration. This study investigated the developmental patterns and the varying trends observed in speech acoustics with advancement in age and gender. The investigation then contributed in predicting the age of the speakers by analyzing these extracted features using various classification techniques and the result revealed maximum recognition rate by using neuro-fuzzy classifiers. This prediction of age may further help us in analyzing the speech samples in order to predict early speech disorders or language delay in children with neuro-developmental disorder.

A novel optimal fuzzy color image enhancement using particle swarm optimization

This paper presents a new approach to enhance the contrast of color images. The intensity component of Hue, Saturation and Intensity (HSV) color model is fuzzified using Global intensification operator (GINT). A new objective measure called contrast information factor is introduced which is optimized using particle swarm optimization technique to learn the parameters. The enhanced image is evaluated by its entropy, index of fuzziness, contrast information and visual quality factor. Subjective and objective evaluation results clearly show the improvement in the quality of the underexposed images in addition of preserving color and specific image features. Also the shape of the histogram is preserved. The results are also compared with histogram equalization technique.

Test case selection for regression testing of applications using web services based on WSDL specification changes

There is much enthusiasm around web services in todays world. Web Services take the advantage of internet to communicate between two electronic devices connected via a network. Testing a Web Service is a challenge as the Service Requester does not have the source code and somehow needs to fully test the impact of changes on his application. Regression testing verifies the integrity of the application and makes sure that the changes have not introduced new software errors. Our approach involves the parsing of the WSDL XML file to extract information regarding the operation name, input message and output message. Both the original and changed XML files for the web service are parsed to extract their respective information from the port type and message element of WSDL. Then, we generate a hash table form the extracted information for both the original and delta WSDL. We pass the hash tables to a Comparator as input, which then compares the hash tables and generates the operation changes as output. In the last step test cases are selected for regressing testing of the changed web service based upon the changes in operations provided by the comparator.

Classification of the Speech of Normally Developing and Intellectually Disabled Children

Development of speech is very crucial for good quality of life. People with learning disabilities face challenges in communication due to deficiencies in speech production and require interventions in terms of training and rehabilitation to become less dependent. For providing specific trainings, it will be valuable to know the exact nature of these deficiencies through acoustic analysis. This study investigated fundamental frequency and intensity in a speech of 95 subjects including 39 normal developing children, 20 adults and 36 subjects with mild to moderate intellectual disability (mental retardation). The results show significant differences in these acoustic measurements. This present study also developed a model that classify the normal developing and intellectually disable groups based on these acoustic features. The findings suggest that acoustic cues such as fundamental frequency along with intensity play a significant role in classifying the groups. It was shown that classifiers with good accuracy can be built based on these parameters which indicate differentiating capabilities of the said features. Such attempts to build classification models can also aid in early diagnosis of intellectual or learning disabilities.

Comparative Study of Pitch Contour for Mentally Impaired

Basic needs of humans are met by conveying information to each other. Most efficient way to do this is via speech signal. But looking in the society we find many people with mental impairment, their essential infrastructure of articulation, word stress, accent, pitch and other speech features are deeply affected by damage in their brain. There are many factors that affect speech but the fundamental frequency (perceived form: pitch) plays major role in prosodic development. This study is an attempt to observe the pattern of pitch development in normal developing and mentally retarded children and adults. It also investigates possible differences between them. Results reveal that younger children have a slighter difference in their pitch as compared to age matched MR (Mentally Retard), but significant difference is found between normal developing adults and those suffering from neurodevelopment disorder. These criteria can be considered as first step in diagnosis of these neurodevelopment disorders.

Age Classification with LPCC Features Using SVM and ANN

For humans, speech is one of the vital communication channel used for interchanging information, knowledge, and thoughts. Identifying the age of a person based on his/her speech is an essential part of speech therapy and many telecommunication applications. Many speech-related disorders can be diagnosed and cured using age identification at early ages. Depending on the age group, particular speech therapy can be given to a child. In this research, typical speech sentences were used to identify the age of 200 Indian children from the age group of 4–8 years. Linear predictive cepstral coefficients (LPCC) (formant frequencies) was applied to extract 128 acoustic features using sustained phonation, reading and imitation tasks. Artificial neural network (ANN) and support vector machine (SVM) were used to build two classification models. Comparisons were made on classification accuracy. Classification results were substantially higher between the age group of 4 and 8 years. This work will further be extended to gender classification with more robust features and algorithms.

Development of spectro-temporal features of speech in children

Spectro-temporal features of speech are the basis of phonological comprehension and production in the brain. Thus, these features provide a relevant framework to study speech and language development in children. In this paper, we present a novel framework to study the statistics of spectro-temporal features of speech that are encoded at different timescales. These timescales correspond to different linguistic units such as prosodic or syllabic components. The framework is tested on a speech corpus consisting of 169 speech samples. The paper demonstrates usage of the proposed framework in finding milestones of speech development in children. The results indicate the presence of more number of spectro-temporal features encoded at short timescales in adults as compared to children. However, no significant difference is observed in the spectro-temporal features encoded at long timescales between these groups. The proposed framework is also used in studying the speech impairments of children and adults with mild to moderate intellectual disabilities. The results reveal the absence of some spectro-temporal features encoded at both the timescales and their absence is more prominent in shorter timescales. The suggested framework can be used for studying speech development and impairment in different disorders.

Developmental changes of spectral parameter in children speech

The objective of this study is to analyze the developmental changes in the speech of children and adult. We quantified speech in terms of the mean power across 10 frequency bins between 0-2000 Hz in the population of 90 children (4-8 years old) and 27 adults. We found that with age, the harmonic structures in the speech increases, as the power spectra of adult is more distributed and has peaks at higher frequency bins than the younger children. To further support our findings, we have performed classification by different classification techniques and found the significant differences in the power spectras of children and adults. This result can be used as a database for Indian children for doing comparative studies to find out speech abnormalities in the children with speech, language or brain disorder.