Handwritten Text Recognition using Deep Learning and Word Beam Search

This paper offers a solution to traditional handwriting recognition techniques using concepts of Deep learning and Word Beam Search. This paper explains about how an individual handwritten word is classified from the handwritten text by translating into a digital form. The digital form when trained with the Connectionist Temporal Classification (CTC) loss function, the output produced is a RNN. This is a matrix containing character probabilities for each time-step. The final text is mapped using a CTC decoding algorithm by converting the character probabilities. The recognized text is constructed by a list of words from the dictionary by using the token passing algorithm. It is found the running time of token passing depends on the size of dictionary. Also the numbers like arbitrary character strings will not able to decode. In this paper the decoding search algorithm word beam search is proposed, in order to tackle these types of problems. This methodology support to constrain words similar to those contained in a dictionary. It allows the character strings such as arbitrary non-word between the words, and integrates into a word-level language model. It is found the running time is better when compared with the token passing. The proposed algorithm comprises of the decoding algorithm named vanilla beam search and token passing using the IAM dataset and Bentham data set.

Energy Harvesting in Cognitive Networks Using Feasible Channel Selection Assignment

Background: The importance of this paper is to achieve maximum spectrum efficiency and proper channel allotment between Primary and Secondary User. The licensed and unlicensed users gets promoted as the channel allotment is properly carried out. To improve energy capability and spectral proficiency consider energy collecting cognitive radio systems to update both energy feasibility and spectral viability. Energy Harvesting Provides possibility of sharing energy in wireless networks which improves the performance of channel capacity. Methods: In this paper an Token Passing algorithm is proposed that switches the channels between Primary User and Secondary User. The energy efficiency decision is taken according to when primary user is idle or not. When the primary user is idle the secondary user cannot harvest any energy and when the primary channel is occupied the secondary channel harvest energy from primary user so that the harvested energy will be used by the secondary user during channel allotment. This proposed algorithm provides energy harvesting and spectrum efficiency. Results: The result shows that the most extraordinary achievable throughput R (eh) of the energy harvesting cognitive radio. The State Transition will move from busy to idle and idle to busy which is represented as S0 and S1. The other parameters are Sensing Energy es, Sampling frequency fs, Primary Signal which accepts a noise SNR γp. As Token Passing Algorithm provides tokens for primary and secondary user it takes lesser time and achieves better throughput than the FDMA and suboptimal algorithm. Conclusion: This paper achieves the maximum spectrum efficiency and energy harvesting by properly allotting spectrum for both primary and secondary user. The primary user and secondary user and spectrum management perform the channel allotment efficiently through the idle and busy state and Token Passing Algorithm does energy harvesting. An efficient scheme is developed for allocating energy in energy harvesting cognitive radio systems.

Automatic Speech Recognition (ASR) System for Isolated Marathi Words: using HTK

The present manuscript focuses on building automatic speech recognition (ASR) system for Marathi language (M-ASR) using Hidden Markov Model Toolkit (HTK). The M-ASR system gives the detail about experimentation and implementation using the HTK Toolkit. In this work total 106 speaker independent Marathi isolated words were recognized. These unique Marathi words are used to train and evaluate M-ASR system. The speech corpus (database) is created by us using isolated Marathi words uttered with mixed gender people. The system uses Mel Frequency Cepstral Coefficient (MFCC) for the purpose of extracting features using Gaussian mixture model (GMM). Viterbi algorithm based on token passing is used for decoding to recognize unknown utterances. The proposed M-ASR system is speaker independent. The proposed system has reported 96.23% word level recognition accuracy.

A Distributed Token Passing Protocol for Time Constrained Data Gathering in VANETs

This paper proposes a novel approach for time constrained information gathering in a typical Vehicular Ad Hoc Network (VANET), based on a token passing scheme, adapted to wireless communications by creating a virtual ring where nodes are connected to a predecessor and a successor node. To address the typical fast topology changes of VANETs, we proposed a specific approach, called Tom Thumb that is a distributed protocol that node-by-node circulates a special packet, called token, which collects the information stored in each vehicle until returning to the first unit within a specified time constraint. The protocol has been properly designed in terms of (i) the more effective hop-by-hop and distributed heuristic implementing the objective function (ii) the token packet format, i.e., the syntax and semantics of its fields. Finally, the performance of the proposed approach is validated for different time constraints and numbers of vehicles, always pointing out a remarkable gain, especially in the presence of severe constraints, i.e., in terms of time deadline, collected information amount and success probability.

Statistical Language Processing for Text Entry

In this chapter we explain how methods from statistical language processing serve as a foundation for the design of probabilistic text entry methods and error correction methods. We review concepts from information theory and language modelling and explain how to design a statistical decoder for text entry—a generative probabilistic model based on the token-passing paradigm. We then present five example applications of statistical language processing for text entry: correcting typing mistakes, enabling fast typing on a smartwatch, improving prediction in augmentative and alternative communication, enabling dwell-free eye-typing and intelligently supporting error correction of probabilistic text entry. We then discuss the limitations of the models presented in this chapter and highlight the importance of establishing solution principles based on engineering science and empirical research in order to guide the design of probabilistic text entry.