Brejesh Lall

Improving Collaborative Filtering Based Recommenders Using Topic Modelling

Standard Collaborative Filtering (CF) algorithms make use of interactions between users and items in the form of implicit or explicit ratings alone for generating recommendations. Similarity among users or items is calculated purely based on rating overlap in this case, without considering explicit properties of users or items involved, limiting their applicability in domains with very sparse rating spaces. In many domains such as movies, news or electronic commerce recommenders, considerable contextual data in text form describing item properties is available along with the rating data, which could be utilized to improve recommendation quality. In this paper, we propose a novel approach to improve standard CF based recommenders by utilizing latent Dirichlet allocation (LDA) to learn latent properties of items, expressed in terms of topic proportions, derived from their textual description. We infer users topic preferences or user profile in the same latent space, based on her historical ratings. While computing similarity between users, we make use of a combined similarity measure involving rating overlap as well as similarity in the latent topic space. This approach alleviates sparsity problem as it allows calculation of similarity between users even if they have not rated any items in common. Our experiments on multiple public datasets indicate that the proposed hybrid approach significantly outperforms standard User Based and Item Based CF recommenders in terms of classification accuracy metrics such as precision, recall and F-measure.

Perceptual Depth Preserving Saliency based Image Compression

The paper tackles the task of preserving depth cue in single images for human perception of depth despite compression artifact. The process is mainly divided into 2 steps; saliency map computation and image compression on the basis of salient regions. The idea is that the regions which are less blurred should be more salient rather than blurred regions. For saliency computation the proposed method first obtains the blur information for the image using frequency information. Then we generate saliency map by region combining on the basis of this blur index. These saliency values are used to vary the quality parameter over the blocks in the image for JPEG compression. Experimental results show that the proposed scheme yields good results over normal JPEG compresssion in terms of PSNR values.

Development of a microcontroller based electrical impedance tomography system

Electrical Impedance Tomography (EIT) is one of the growing and optimistic imaging modality for clinical imaging with such advantages as: non-invasive, low cost, ease of use and practicality, this method has been used in many areas of science and research. This paper consists of designing and development of hardware for EIT system using microcontroller MSP430G2 and AVR ATMEGA 16. A 16-electrode phantom is developed and studied with an instrumentation for EIT. Using serial communication data applies to PC for image reconstruction using EIDORS.

Issues in the Implementation of General Dual-Rate Systems

General dual-rate systems enhance the design free- dom by removing the structural dependencies and hence can be used in the incompatible nonuniform filter banks. Recently it was shown that realising these dual-rate systems using a simple linear time invariant (LTI) system, a conventional sampler and a block sampler, can reduce the circuit layout complexity. The dual- rate system specifications are obtained in the Multi Input Multi Output (MIMO) representation, in the design of the filter bank. This paper address the issue of designing the LTI system for the simple realisations from the MIMO specifications. We obtain the relationship between the polyphase decomposition of the LTI system and the transfer matrix of the MIMO representation. Our analysis show that the simple realisations are obtained from generalising the switching representations of linear periodically time varying (LPTV) systems. Index Terms—Multirate signal processing, MIMO representa- tion, general dual-rate systems. systems required in the proposed implementations. The one- one correspondence proved in (2), between the time indices is not useful in designing the required LTI systems. So in this paper we address the problem of relating the transfer matrix of MIMO representation to the transfer func- tion of the required LTI. The conventional samplers in the MIMO representations are moved from one side to the other using the Noble identities (7). Analysing the resulting systems revealed the relationship between the rows and columns of the transfer matrix to the polyphase decomposition of the required LTI systems. From our analysis the realisations in (2) were found to be a generalisation of the switching representations (4) of LPTV systems. This paper is organised as follows. Section-II reviews the linear dual-rate systems and their MIMO representations. In section-III we present the design advantage of general dual- rate structures in incompatible filter banks. Section-IV derives the relationship between the MIMO representation and the LTI systems in the proposed realisations. Section-V contains the conclusions. A. Notations

Alias reduction: Generalised pseudocirculant conditions

Complete alias cancellation, in an arbitrary nonuniform and/or nonmaximally decimated filter bank structure, is not guaranteed. Also it is well known that the inversion of a linear periodically time varying (LPTV) system is not always guaranteed. In this paper we exploit the ability of uniform filter bank (UFB) framework to provide alternative models for these structures. We propose a simplification by generalising the well known pseudocirculant conditions [1] (Vaidyanathan and Mitra, 1988). UFB can be characterised by the two switching representations of the LPTV systems. The set of linear time invariant (LTI) systems in each representation is uniquely related to the rows and columns of the matrix associated with the UFB. Under the pseudocirculant conditions the UFB reduces to an LTI system. Here we obtain a set of generalised pseudocirculant conditions which reduce the UFB to an LPTV of lower order.

Detecting and Correlating Video-Based Event Patterns: An Ontology Driven Approach

With increasing amount of information (video, text) being available today, it has become non-trivial to develop techniques to categorize documents into contextually meaningful classes. The information as available in the documents is composed of sequence of events termed as patterns. It is evident to know the important trends as observed from patterns that are emerging over a specific time period and space. For identifying the patterns, we must focus on semantic meaning of documents. Tracing such patterns in videos or texts manually is a time-consuming, cumbersome or an impossible task. So, in this paper we have devised an unsupervised trend discovery approach that detects and correlates event patterns from videos temporally as well as spatially. We begin by building our own document collection on the basis of contextual meaning of documents. This helps in associating an input video with another video or text documents on the basis of their semantic meaning. This approach helps in accumulating variety of information that is scattered over the web thus providing relatively complete information about the video. The highly correlated words are grouped in a topic using Latent Dirichlet Allocation (LDA). To identify topics an E-MOWL based ontology is used. This event ontology helps in discovering associations and relations between the various events. With this kind of representation, the users can infer different concepts as emerged over time. For identifying the various spatial patterns that exist corresponding to an event in a document, we have utilized geographic ontology (Geoontology). We establish validity of our approach using experimental results.

Saliency map based image steganography

Steganography is an emerging field for hiding secure data. This is useful in electronics data security domains such as watermarking, copyright, secret number, unauthorized reproduction etc. In this paper we propose a saliency based steganography technique, which uses a statistical block analysis of the saliency map to identify regions for secure data embedding. Saliency computation is also an active research area, and for best result, different techniques are combined to generate the saliency map. Various steganography techniques such as sequential, block, and transform domain do not take into account the variation in perceptual information of an image. In saliency based embedding, visual attention is considered for selecting regions for embedding of secure data. The proposed saliency based steganography provides a new dimension for steganography, where embedding of secure data causes minimum affect on visual objects of an image as well as little distortion in segmentation boundaries.

Cyclostationary signals in multirate linear systems

This paper presents a systematic approach to use time frequency representation (TFR) for the analysis of cyclostationary signals in multirate linear systems. We exploit the ability of blocking operation to perform the analysis in a simple yet efficient manner. We illustrate the strength of TFR for performing such analysis. The basic idea is — TFR of a cyclostationary signal is directly related to the rows of the power spectrum matrix of its blocked version. We present examples including nonuniform filter bank to highlight the capabilities of the analysis.

Practical phantom study of low cost portable EIT based cancer screening device

Electrical Impedance Tomography (EIT) is one of the most potential and promising imaging technique which has wide applications in the medical field. This paper consists of practical phantom study conducted during the designing and development of the Arduino Uno based hardware of a low cost, portable cancer screening device. In this study a phantom tank with 16 Stainless Steel electrodes is developed along with various blocks (VCO,VCCS and MUX) for testing and studying various organic phantoms.

IIR deconvolution from noisy observations using Kalman filtering

In this paper, we reconstruct the input signal of an IIR filter from the noise corrupted output signal. We perform two operations parallely. One deconvolution and the other, noise removal. We show how to use Kalman filter to perform this task. We develop theory for a very general scenario of reconstructing an ARMA process from its noise corrupted IIR filtered output. We develop augmented state space equations combining the state space equations of the ARMA process and the IIR filter, which are required to apply Kalman filter. The simulation results show clear improvement in the signal-to-noise ratio.