Somasheker Akkaladevi

Improving multiple-password recall: an empirical study

As one of the most common authentication methods, passwords help secure information by granting access only to authorized parties. To be effective, passwords should be strong, secret, and memorable. While password strength can be enforced by automated information technology policies, users frequently jeopardize secrecy to improve memorability. The password memorability problem is exacerbated by the number of different passwords a user is required to remember. While short-term memory theories have been applied to individual-password management problems, the relationship between memory and the multiple-password problem has not been examined. This paper treats the multiple-password management crisis as a search and retrieval problem involving human beings’ long-term memory. We propose that interference between different passwords is one of the major challenges to multiple-password recall and that interference alleviation methods can significantly improve multiple-password recall. A lab experiment was conducted to examine the effectiveness of two interference alleviation methods: the list reduction method and the unique identifier method. While both methods improve multiple-password recall performance, the list reduction method leads to statistically significant improvement. The results demonstrate the potential merit of practices targeting multiple-password interference. By introducing long-term memory theory to multiple-password memorability issues, this study presents implications benefiting users and serves as the potential starting point for future research.

Granular neural web agents for stock prediction

Abstract A granular neural Web-based stock prediction agent is developed using the granular neural network (GNN) that can discover fuzzy rules. Stock data sets are downloaded from www.yahoo.com website. These data sets are inserted into the database tables using a java program. Then, the GNN is trained using sample data for any stock. After learning from the past stock data, the GNN is able to use discover fuzzy rules to make future predictions. After doing simulations with six different stocks (msft, orcl, dow, csco, ibm, km), it is conclusive that the granular neural stock prediction agent is giving less average errors with large amount of past training data and high average errors in case of fewer amounts of past training data. Java Servlets, Java Script and jdbc are used. SQL is used as the back-end database. The performance of the GNN algorithm is compared with the performance of the BP algorithm by training the same set of data and predicting the future stock values. The average error of the GNN is less than that of BP algorithm.

Protein secondary structure prediction using neural network and simulated annealing algorithm

Predicting the secondary structure of a protein (alpha-helix, beta-sheet, coil) is an important step towards elucidating its three dimensional structure, as well as its function. In this research we use a multilayer feed forward neural network for protein secondary structure prediction. The RS126 data set was used for training and testing the proposed neural network. We combined neural network and simulated annealing (SA) to further improve on the accuracy of protein secondary structure prediction. The results obtained show that by combining the neural network with SA technique improves the prediction accuracy in the range of 2-3%.

Protein Secondary Structure Prediction using Bayesian Inference method on Decision fusion algorithms

Prediction of protein secondary structure (alpha-helix, beta-sheet, coil) from primary sequence of amino acids is a very challenging task, and the problem has been approached from several angles. Previously research was performed in this field using several techniques such as neural networks, simulated annealing (SA) and genetic algorithms (GA) for improving the protein secondary structure prediction accuracy. Decision fusion methods such as the committee method and correlation methods were also used in combination with the profile-based neural networks and AI algorithms for achieving better prediction accuracy. In this research we investigate the Bayesian inference method for predicting the protein secondary structure. The Bayesian inference method proposed in this research uses the results from the committee and correlation methods to achieve better prediction accuracy. Simulations are performed using the RS126 data set. The results show that the protein secondary structure prediction accuracy can be improved by more than 2% using the Bayesian inference method.

Message Routing and Scheduling in Optical Multistage Networks using Bayesian Inference method on AI algorithms

Optical multistage interconnection networks (MINs) suffer from optical-loss during switching and crosstalk problem in the switches. The crosstalk problem is solved by routing messages using time division multiplexing (TDM) approach. This paper focuses on minimizing the number of groups (time slots) required to realize a permutation. Many researchers concentrated on this NP-hard problem and concluded that AI algorithms perform better than the heuristic algorithms. They also showed that majority of the times the performance of genetic algorithm (GA) was better than simulated annealing algorithm (SAA). In this research, we implement a new approach to minimize the number of passes required for scheduling a given permutation. A combinational method is developed which comprises the use of Bayesian inference method on GA and SAA to always guarantee the best solution, instead of only using either GA or SAA. Simulations are performed in Java using multiple threads to run SA and GAA in parallel and to evaluate the performance of the new method. The results are then compared to those obtained from GA and SAA.

Strip and line discrete cosine transforms for blocking artifact reduction

The standard two dimensional block discrete cosine transform suffers greatly from the blocking artifacts when used in compression. A modified block assignment is used to reduce the block discontinuity associated with the standard DCT. The new block assignment spans a strip of several rows or columns. Several sizes of these blocks have been implemented, the smallest is a single line which reduces the two dimensional DCT into one dimension. Both the two and one dimensional strip discrete cosine transforms were used in this paper to achieve compression with considerable reduction in blocking artifacts.