V Srinivasa Chakravarthy

We present Bharati, a simple, novel script that can represent the characters of a majority of contemporary Indian scripts. The shapes/motifs of Bharati characters are drawn from some of the simplest characters of existing Indian scripts. Bharati characters are designed such that they strictly reflect the underlying phonetic organization, thereby attributing to the script qualities of simplicity, familiarity, ease of acquisition and use. Thus, employing Bharati script as a common script for a majority of Indian languages can ameliorate several existing communication bottlenecks in India. We perform a complexity analysis of handwritten Bharati script and compare its complexity with that of nine major Indian scripts. The measures of complexity are derived from a theory of handwritten characters based on Catastrophe theory. Bharati script is shown to be simpler than the nine major Indian scripts in most measures of complexity. Self-organizing maps (SOM) are generated by training data for handwritten characters of Bharati and Indian scripts. Phonetically similar characters are clustered together on SOM for Bharati, supporting the proposition that the shapes of Bharati script follow the underlying phonetic organization.

Deep Brain Stimulation (DBS) of Sub Thalamic Nucleus (STN) is the most sought out therapeutic technique for the treatment of motor symptoms in advanced parkinsonian conditions. But the effect of STN-DBS on cognition was observed to be contrary with Impulsivity being observed as the most common side effect. Among the numerous behavioral tasks, Iowa Gambling Task (IGT) captures one of the impulsivity features (premeditation) and the task resembles real life decision making scenario. A 2D spiking network of basal ganglia (BG) was modeled to study the cognitive aspects of Parkinson's disease (PD) during medication and DBS. The model consist of key BG nuclei such as the Globus Pallidus externus (GPe) and Globus Pallidus internus (GPi) and STN modeled as Izhikevich 2D spiking neurons and striatal output as Poisson process. The concept of dopamine being the reward prediction error was utilized to update the cortico-striatal weights. The model was then tested on 3 conditions i.e., healthy controls, PD 'ON' and STN-DBS. The effect of DBS on decision making (in terms of IGT score) was studied by changing the electrode position (in STN) in the model. Our results indicate that changing the electrode's position and current spread independently leads to a critical change in performance levels. The model also shows that simulated PD 'ON' medication performed poorly compared to healthy controls as observed in experiments. The simulated results suggest that electrode position or current spread might be the probable reason for the observed controversial outcomes in STN-DBS patients. This is one of the first models to use spiking neurons to test the effects of dopamine medications and STN DBS on complex decision making.

In this paper, we address some issues in developing an online handwritten character recognition(HCR) system for an Indian language script, Telugu. The number of characters in this script is estimated to be around 5000. A character in this script is written as a sequence of strokes. The set of strokes in Telugu consists of 253 unique strokes. As the similarity among several strokes is high, we propose a modular approach for recognition of strokes. Based on the relative position of a stroke in a character, the stroke set has been divided into three subsets, namely, baseline strokes, bottom strokes and top strokes. Classifiers for the different subsets of strokes are built using support vector machines(SVMs). We study the performance of the classifiers for subsets of strokes and propose methods to improve their performance. A comparative study using hidden Markov models(HMMs) shows that the SVM based approach gives a significantly better performance. © 2007 IEEE.

Saccades are rapid, frequent eye movements that shift the fovea onto objects of interest. Several areas of the brain, including the frontal cortical areas, Lateral Intraparietal (LIP) cortex, Basal Ganglia (BG), Superior Colliculus (SC) and the brainstem reticular formation are believed to be involved in saccade generation. Models of saccade generation, however, tend to focus heavily on the determination of saccadic saliency in the Superior Colliculus and eye movement dynamics thereafter. The aim of this paper is to model the role played by the BG, particularly the switching function of the SubThalamic Nucleus (STN), in switching from a pattern of automated saccadic responses, to voluntary control of saccade generation. © 2010 IEEE.

A system for online recognition of handwritten Tamil characters is presented. A handwritten character is constructed by executing a sequence of strokes. A structure- or shape-based representation of a stroke is used in which a stroke is represented as a string of shape features. Using this string representation, an unknown stroke is identified by comparing it with a database of strokes using a flexible string matching procedure. A full character is recognized by identifying all the component strokes. Character termination, is determined using a finite state automaton. Development of similar systems for other Indian scripts is outlined. © 2004 IEEE.

We combine the strengths of four different pattern analysis techniques to develop a powerful and efficient system for handwritten character recognition. The four techniques are: 1) Convolutional neural networks (CNN), 2) Principal Component Analysis (PCA), 3) Support vector machines, 4) Multiclassifier systems. The proposed system that embodies the above-mentioned four techniques is used for recognition of offline handwritten Telugu characters. Telugu aksharas of consonant-vowel (CV) type, with 36 consonant classes and 15 vowel modifier classes, are used for the study. Telugu dataset consisted of 47428 CV images in the training set and 5156 CV images in the test set. In addition to Telugu dataset, MNIST database consisting of 60000 digits for training and 10000 digits for testing was used in this study. The proposed system yields a performance of 98.5% on MNIST numeric data, 92.26% and 92% on consonants and vowel modifier of Telugu characters respectively. © 2013 IEEE.

This article proposes an improved XML standard for storing online handwritten data in Indian languages. This standard has evolved over a period of two years, and is currently being used by the Consortium for online handwritten recognition of Indian languages, for annotating about 100,000 handwritten words in each of six Indian languages, namely, Tamil, Kannada, Telugu, Malayalam, Hindi and Bangla. In order that the huge amount of data that is being collected is useable by the future researchers, it is preferable that the data is stored in a format that is unambiguous and easy to read. The uniqueness of this refined standard is that it gives quality labels at different levels to the data, and has provision to annotate all the peculiarities of writing the script of the various Indian languages included in the current consortium project. The current format allows the use of automated and semi-automated annotation tools. Copyright © 2009 ACM.

Area V5 or Middle Temporal (MT) area of the primate brain is said to be involved in visual motion perception. Physiological studies indicate that the neurons in MT respond selectively to the direction of moving stimuli. However in response to the complex stimuli containing multiple oriented components, a set of MT neurons are selective to the direction of the component motion whereas the other set of MT neurons are selectively respond to the direction of the whole pattern motion. This paper discusses a two layer LISSOM model (Laterally Interconnected Synergetically Self-Organizing Map) which is analogous to neurons in the cortical areas V1 as well as MT. The adaptive Hebbian learning technique has been used to train the network with sequences of moving square stimuli and observed the following: i) afferent weight connections of V1 neurons are tuned as orientation detectors and ii) neurons at MT is tuned to the whole pattern motion. Lateral connections in each layer mediate the competition between the neurons which results in a topographic map. The results from the two layer LISSOM model was found to be in-line with that of well known experimental studies results.

We present two schemas for online recognition of Telugu characters, involving elaborate multiclassifier architectures. Considering the three-tier vertical organization of a typical Telugu character, we divide the stroke set into 4 subclasses primarily based on their vertical position. Stroke level recognition is based on a bank of Support Vector Machines (SVMs), with a separate SVM trained on each of these classes. Character recognition for Schema 1 is based on a Ternary Search Tree (TST), while for Schema 2 it is based on a SVM. The two schemas yielded overall stroke recognition performances of 89.59% and 96.69% respectively surpassing some of the recent online recognition performance results related to Telugu script reported in literature. The schemas yield character-level recognition performances of 90.55% and 96.42% respectively. © 2012 IEEE.

Photoplethysmography (PPG) is typically used to extract cardiac-related information like heart rate and cardiac output, though extra-cardiac information like respiratory rate can also be extracted from PPG. The aim of the current study is to advance this approach further and investigate existence of gastric-related activity in PPG. To this end, we consider pulse rate variability (PRV), which provides information analogous to heart rate variability (HRV). Finger PPG and electrogastrography (EGG) signals were recorded from 8 healthy volunteers in fasting and postprandial state for 30 minutes. Peak-to-peak interval (PPI) analysis shows that the power of high frequency (HF) component in fasting and postprandial state changes significantly. The power ratio (PR), which is the ratio between powers of low frequency band (LF, 0.04-0.15 Hz) to that of high frequency band (HF, 0.15-0.4 Hz) and the EGG power were calculated in fasting and postprandial state. PR was positively correlated with EGG power (r = 0.46; P< 0.05). PR indicates the balancing sympathovagal modulation and vagal nervous activity. The significance of this study is that PR from PRV analysis could be used as a tool for diagnosing gastric system instead of the present invasive/cumbersome methods. ©2009 IEEE.