Online Handwritten Kannada Word Recognizer with Unrestricted Vocabulary

In this paper, we propose a novel heuristic approach to segment recognizable symbols from online Kannada word data and perform recognition of the entire word. Two different estimates of first derivative are extracted from the preprocessed stroke groups and used as features for classification. Estimate 2 proved better resulting in 88% accuracy, which is 3% more than that achieved with estimate 1. Classification is performed by statistical dynamic space warping (SDSW) classifier which uses X, Y co-ordinates and their first derivatives as features. Classifier is trained with data from 40 writers. 295 classes are handled covering Kannada aksharas, with Kannada numerals, Indo-Arabic numerals, punctuations and other special symbols like $ and #. Classification accuracies obtained are 88% at the akshara level and 80% at the word level, which shows the scope for further improvement in segmentation algorithm

Laser receptive polyelectrolyte thin films doped with biosynthesized silver nanoparticles for antibacterial coatings and drug delivery applications

We report a simple method to fabricate multifunctional polyelectrolyte thin films to load and deliver the therapeutic drugs. The multilayer thin films were assembled by the electrostatic adsorption of poly (allylamine hydrochloride) (PAH) and dextran sulfate (DS). The silver nanoparticles (Ag NPs) biosynthesized from novel Hybanthus enneaspermus leaf extract as the reducing agent were successfully incorporated into the film. The biosynthesized Ag NPs showed excellent antimicrobial activity against the range of enteropathogens, which could be significantly enhanced when used with commercial antibiotics. The assembled silver nano composite multilayer films showed rupture and deformation when they are exposed to laser. The Ag NPs act as an energy absorption center, locally heat up the film and rupture it under laser treatment. The antibacterial drug, moxifloxacin hydrochloride (MH) was successfully loaded into the multilayer films. The total amount of MH release observed was about 63% which increased to 85% when subjected to laser light exposure. Thus, the polyelectrolyte thin film reported in our study has significant potential in the field of remote activated drug delivery, antibacterial coatings and wound dressings. (C) 2013 Elsevier B.V. All rights reserved.

Recognition of open vocabulary, online handwritten pages in Tamil script

In this work, we describe a system, which recognises open vocabulary, isolated, online handwritten Tamil words and extend it to recognize a paragraph of writing. We explain in detail each step involved in the process: segmentation, preprocessing, feature extraction, classification and bigram-based post-processing. On our database of 45,000 handwritten words obtained through tablet PC, we have obtained symbol level accuracy of 78.5% and 85.3% without and with the usage of post-processing using symbol level language models, respectively. Word level accuracies for the same are 40.1% and 59.6%. A line and word level segmentation strategy is proposed, which gives promising results of 100% line segmentation and 98.1% word segmentation accuracies on our initial trials of 40 handwritten paragraphs. The two modules have been combined to obtain a full-fledged page recognition system for online handwritten Tamil data. To the knowledge of the authors, this is the first ever attempt on recognition of open vocabulary, online handwritten paragraphs in any Indian language.

Predatory and trophobiont-tending ants respond differently to fig and fig wasp volatiles

The interaction between figs and their pollinating or parasitic fig wasps is mediated largely by chemical communication. These fig wasps are often preyed upon by predatory ants. In this study, we found that predatory ants (Oecophylla smaragdina) patrolling Ficus racemosa trees were attracted to the odour from fig syconia at different developmental phases, as well as to the odours of fig wasps, whereas other predatory ants (Technomyrmex albipes) responded only to odours of syconia from which fig wasps were dispersing and to fig wasp odour. However, trophobiont-tending ants (Myrmicaria brunnea) patrolling the same trees and exposed to the same volatiles were unresponsive to fig or fig wasp odours. The predatory ants demonstrated a concentration-dependent response towards volatiles from figs receptive to pollinators and those from which wasps were dispersing while the trophobiont-tending ants were unresponsive to such odours at all concentrations. Naive predatory ants failed to respond to the volatiles to which the experienced predatory ants responded, indicating that the response to fig-related odours is learned. We suggest that predatory ants could use fig-associated volatiles to enhance their probability of wasp encounter and can eavesdrop on signals meant for pollinators. (C) 2009 The Association for the Study of Animal Behaviour.

The chemical ecology of seed dispersal in monoecious and dioecious figs

In the nursery pollination system of figs (Ficus, Moraceae), flower-bearing receptacles called syconia breed pollinating wasps and are units of both pollination and seed dispersal. Pollinators and mammalian seed dispersers are attracted to syconia by volatile organic compounds (VOCs). In monoecious figs, syconia produce both wasps and seeds, while in (gyno)dioecious figs, male (gall) fig trees produce wasps and female (seed) fig trees produce seeds. VOCs were collected using dynamic headspace adsorption methods on freshly collected figs from different trees using Super Q® collection traps. VOC profiles were determined using gas chromatography–mass spectrometry (GC–MS).The VOC profile of receptive and dispersal phase figs were clearly different only in the dioecious mammal-dispersed Ficus hispida but not in dioecious bird-dispersed F. exasperata and monoecious bird-dispersed F. tsjahela. The VOC profile of dispersal phase female figs was clearly different from that of male figs only in F. hispida but not in F. exasperata, as predicted from the phenology of syconium production which only in F. hispida overlaps between male and female trees. Greater difference in VOC profile in F. hispida might ensure preferential removal of seed figs by dispersal agents when gall figs are simultaneously available.The VOC profile of only mammal-dispersed female figs of F. hispida had high levels of fatty acid derivatives such as amyl-acetates and 2-heptanone, while monoterpenes, sesquiterpenes and shikimic acid derivatives were predominant in the other syconial types. A bird- and mammal-repellent compound methyl anthranilate occurred only in gall figs of both dioecious species, as expected, since gall figs containing wasp pollinators should not be consumed by dispersal agents.

Mathematics, nature and cryptography: Insights from philosophy of information

One influential image that is popular among scientists is the view that mathematics is the language of nature. The present article discusses another possible way to approach the relation between mathematics and nature, which is by using the idea of information and the conceptual vocabulary of cryptography. This approach allows us to understand the possibility that secrets of nature need not be written in mathematics and yet mathematics is necessary as a cryptographic key to unlock these secrets. Various advantages of such a view are described in this article.

On Finding the Natural Number of Topics with Latent Dirichlet Allocation: Some Observations

It is important to identify the ``correct'' number of topics in mechanisms like Latent Dirichlet Allocation(LDA) as they determine the quality of features that are presented as features for classifiers like SVM. In this work we propose a measure to identify the correct number of topics and offer empirical evidence in its favor in terms of classification accuracy and the number of topics that are naturally present in the corpus. We show the merit of the measure by applying it on real-world as well as synthetic data sets(both text and images). In proposing this measure, we view LDA as a matrix factorization mechanism, wherein a given corpus C is split into two matrix factors M-1 and M-2 as given by C-d*w = M1(d*t) x Q(t*w).Where d is the number of documents present in the corpus anti w is the size of the vocabulary. The quality of the split depends on ``t'', the right number of topics chosen. The measure is computed in terms of symmetric KL-Divergence of salient distributions that are derived from these matrix factors. We observe that the divergence values are higher for non-optimal number of topics - this is shown by a `dip' at the right value for `t'.

Measurement of complex optical flow with use of an augmented generalized gradient scheme

We present a method for measuring the local velocities and first-order variations in velocities in a timevarying image. The scheme is an extension of the generalized gradient model that encompasses the local variation of velocity within a local patch of the image. Motion within a patch is analyzed in parallel by 42 different spatiotemporal filters derived from 6 linearly independent spatiotemporal kernels. No constraints are imposed on the image structure, and there is no need for smoothness constraints on the velocity field. The aperture problem does not arise so long as there is some two-dimensional structure in the patch being analyzed. Among the advantages of the scheme is that there is no requirement to calculate second or higher derivatives of the image function. This makes the scheme robust in the presence of noise. The spatiotemporal kernels are of simple form, involving Gaussian functions, and are biologically plausible receptive fields. The validity of the scheme is demonstrated by application to both synthetic and real video images sequences and by direct comparison with another recently published scheme Biol. Cybern. 63, 185 (1990)] for the measurement of complex optical flow.

Measurement of complex optical flow with use of an augmented generalized gradient scheme

We present a method for measuring the local velocities and first-order variations in velocities in a time-varying image. The scheme is an extension of the generalized gradient model that encompasses the local variation of velocity within a local patch of the image. Motion within a patch is analyzed in parallel by 42 different spatiotemporal filters derived from 6 linearly independent spatiotemporal kernels. No constraints are imposed on the image structure, and there is no need for smoothness constraints on the velocity field. The aperture problem does not arise so long as there is some two-dimensional structure in the patch being analyzed. Among the advantages of the scheme is that there is no requirement to calculate second or higher derivatives of the image function. This makes the scheme robust in the presence of noise. The spatiotemporal kernels are of simple form, involving Gaussian functions, and are biologically plausible receptive fields. The validity of the scheme is demonstrated by application to both synthetic and real video images sequences and by direct comparison with another recently published scheme [Biol. Cybern. 63, 185 (1990)] for the measurement of complex optical flow.

Unrestricted Kannada online handwritten akshara recognition using SDTW

In this paper, we present an unrestricted Kannada online handwritten character recognizer which is viable for real time applications. It handles Kannada and Indo-Arabic numerals, punctuation marks and special symbols like $, &, # etc, apart from all the aksharas of the Kannada script. The dataset used has handwriting of 69 people from four different locations, making the recognition writer independent. It was found that for the DTW classifier, using smoothed first derivatives as features, enhanced the performance to 89% as compared to preprocessed co-ordinates which gave 85%, but was too inefficient in terms of time. To overcome this, we used Statistical Dynamic Time Warping (SDTW) and achieved 46 times faster classification with comparable accuracy i.e. 88%, making it fast enough for practical applications. The accuracies reported are raw symbol recognition results from the classifier. Thus, there is good scope of improvement in actual applications. Where domain constraints such as fixed vocabulary, language models and post processing can be employed. A working demo is also available on tablet PC for recognition of Kannada words.

Kannada WordNet - a lexical database

This paper presents the preliminary analysis of Kannada WordNet and the set of relevant computational tools. Although the design has been inspired by the famous English WordNet, and to certain extent, by the Hindi WordNet, the unique features of Kannada WordNet are graded antonyms and meronymy relationships, nominal as well as verbal compoundings, complex verb constructions and efficient underlying database design (designed to handle storage and display of Kannada unicode characters). Kannada WordNet would not only add to the sparse collection of machine-readable Kannada dictionaries, but also will give new insights into the Kannada vocabulary. It provides sufficient interface for applications involved in Kannada machine translation, spell checker and semantic analyser.

Tuned Normalization Explains the Size of Attention Modulations

The effect of attention on firing rates varies considerably within a single cortical area. The firing rate of some neurons is greatly modulated by attention while others are hardly affected. The reason for this variability across neurons is unknown. We found that the variability in attention modulation across neurons in area MT of macaques can be well explained by variability in the strength of tuned normalization across neurons. The presence of tuned normalization also explains a striking asymmetry in attention effects within neurons: when two stimuli are in a neuron's receptive field, directing attention to the preferred stimulus modulates firing rates more than directing attention to the nonpreferred stimulus. These findings show that much of the neuron-to-neuron variability in modulation of responses by attention depends on variability in the way the neurons process multiple stimuli, rather than differences in the influence of top-down signals related to attention.

Fast Likelihood Computation in Speech Recognition using Matrices

Acoustic modeling using mixtures of multivariate Gaussians is the prevalent approach for many speech processing problems. Computing likelihoods against a large set of Gaussians is required as a part of many speech processing systems and it is the computationally dominant phase for Large Vocabulary Continuous Speech Recognition (LVCSR) systems. We express the likelihood computation as a multiplication of matrices representing augmented feature vectors and Gaussian parameters. The computational gain of this approach over traditional methods is by exploiting the structure of these matrices and efficient implementation of their multiplication. In particular, we explore direct low-rank approximation of the Gaussian parameter matrix and indirect derivation of low-rank factors of the Gaussian parameter matrix by optimum approximation of the likelihood matrix. We show that both the methods lead to similar speedups but the latter leads to far lesser impact on the recognition accuracy. Experiments on 1,138 work vocabulary RM1 task and 6,224 word vocabulary TIMIT task using Sphinx 3.7 system show that, for a typical case the matrix multiplication based approach leads to overall speedup of 46 % on RM1 task and 115 % for TIMIT task. Our low-rank approximation methods provide a way for trading off recognition accuracy for a further increase in computational performance extending overall speedups up to 61 % for RM1 and 119 % for TIMIT for an increase of word error rate (WER) from 3.2 to 3.5 % for RM1 and for no increase in WER for TIMIT. We also express pairwise Euclidean distance computation phase in Dynamic Time Warping (DTW) in terms of matrix multiplication leading to saving of approximately of computational operations. In our experiments using efficient implementation of matrix multiplication, this leads to a speedup of 5.6 in computing the pairwise Euclidean distances and overall speedup up to 3.25 for DTW.

Supervised matching of comments with news article segments

Comments constitute an important part of Web 2.0. In this paper, we consider comments on news articles. To simplify the task of relating the comment content to the article content the comments are about, we propose the idea of showing comments alongside article segments and explore automatic mapping of comments to article segments. This task is challenging because of the vocabulary mismatch between the articles and the comments. We present supervised and unsupervised techniques for aligning comments to segments the of article the comments are about. More specifically, we provide a novel formulation of supervised alignment problem using the framework of structured classification. Our experimental results show that structured classification model performs better than unsupervised matching and binary classification model.

Fast acoustic likelihood computation using low-rank matrix approximation

Acoustic modeling using mixtures of multivariate Gaussians is the prevalent approach for many speech processing problems. Computing likelihoods against a large set of Gaussians is required as a part of many speech processing systems and it is the computationally dominant phase for LVCSR systems. We express the likelihood computation as a multiplication of matrices representing augmented feature vectors and Gaussian parameters. The computational gain of this approach over traditional methods is by exploiting the structure of these matrices and efficient implementation of their multiplication.In particular, we explore direct low-rank approximation of the Gaussian parameter matrix and indirect derivation of low-rank factors of the Gaussian parameter matrix by optimum approximation of the likelihood matrix. We show that both the methods lead to similar speedups but the latter leads to far lesser impact on the recognition accuracy. Experiments on a 1138 word vocabulary RM1 task using Sphinx 3.7 system show that, for a typical case the matrix multiplication approach leads to overall speedup of 46%. Both the low-rank approximation methods increase the speedup to around 60%, with the former method increasing the word error rate (WER) from 3.2% to 6.6%, while the latter increases the WER from 3.2% to 3.5%.