690 research outputs found
Written evidence submitted by Dr Deepak Padmanabhan (Sr Lecturer at Queen's University Belfast) [ National Security Strategy (Joint Committee) > Defending Democracy]
No full textWritten evidence submitted by Dr Deepak Padmanabhan (Sr Lecturer at Queen's University Belfast) [ National Security Strategy (Joint Committee) > Defending Democracy]
No full textRecommended from our members
Relational Care - with Mary Larkin and Manik Deepak-Gopinath [Podcast]
No full textWhat is 'relational care' and how can it improve the day-to-day experience of carers and those they care for? What are its implications for relationships between staff and service users in care settings? And how does the concept of relational care enable us to re-imagine the role of place and space in the experience of care? These are some of the questions we explore in this episode with Mary Larkin and Manik Deepak-Gopinath who recently completed a research project on the value and practice of relational care with older people.
Mary is Professor of Care, Carers and Caring at The Open University in the UK, where her research has focused on carers and caring and adult social care. She is the co-author, most recently of Family Carers and Caring, published in 2023 by Emerald. Manik is a Lecturer in Ageing, also at The Open University, and is a critical gerontologist with interests in the intersection of ageing, place and wellbeing, and in the intimate and family ties of older adults
My Name Is Deepak
No full textThis chapter looks at the author's responses to being given a nickname by his co-workers: Tupac. They do it in a friendly manner, but the author doesn’t understand the connection with the American rapper. It makes him think about who he is, his identity, and how people see him in his adopted country.</p
Sideffective - system to mine patient reviews: sentiment analysis
No full textSideffective is the system to crawl, rank and analyze patient testimonials about side ffeects from common medications. Since the wealth of any mining model is the Data corpus, the data collection phase involved extensive crawling of massive medical websites comprised of user forums from the internet. Subsequently, the raw files were subjected to certain site-specific parsing routines, yielding outputs conforming to a well-defined data model. Currently, the system holds close to 400,000 user testimonials pertaining to more than 2500 drugs/medicines. Sideffective aims at gathering and aggregating this wealth of information, build useful associations and present interesting observations and numeric validations, all in a user-friendly interface. The important issues that we have tried to tackle are: Extracting side effects without relying on pre-built lists, aggregating distribution of different side effect for a give drug, site-specific search, ranking and determining the negativity of reviews. The system has been jointly built by Deepak Yalamanchi and Sangeetha Rajagopalan under the guidance of Prof. Tomasz Imielinski. This thesis focuses mainly on Sentiment Analysis of patient reviews. While most existing sentiment analysis systems are predicated by POS (parts of speech) tagging or Bayesian sentiment analysis methods, the same cannot be applied to medical reviews as they generally carry a negative flavor in them. We thereby approached the problem by identifying the features in the sentence and calibrating the sentiment on a Negativity Meter based on their relation to sentiment words. A feature, as defined for the purpose of this thesis, can be a medicine, a side effect or a symptom. The sentiment of each feature is determined by the aggregate of all its polarities with respect to each sentiment word, where the polarity is determined by an inverse relation to the distance of the feature from the sentiment word. Each sentence is then evaluated by the cumulative polarity of all the features contained in it. Sentiment of a review is determined by individually determining the sentiment of each sentence and then getting a weighted sum score of all the sentences in the review. The accuracy of a sentiment analysis system is, in principle, how well it agrees with human judgments. Experimental results, involving human reviewers (extracted from site: www.askapatient.com) and correlating them back to the negativity rating of each review yield conclusive results, demonstrating the effectiveness of the technique. We have also implemented a customized Lucene search on the data using a multi-review summarization approach and a ranking scheme based on the feature-list. Ranking priority is given to the review that has the largest feature list size.M.S.Includes bibliographical referencesby Deepak Yalamanch
Multi-View Clustering
No full textWith a plethora of data capturing modalities becoming available, the same data object often leaves different kinds of digital footprints. This naturally leads to datasets comprising the same set of data objects represented in different forms, called multi-view data. Among the most fundamental tasks in unsupervised learning is that of clustering, the task of grouping data objects into groups of related objects. Multi-view clustering (MVC) is a flourishing field in unsupervised learning; the MVC task considers leveraging multiple views of data objects in order to arrive at a more effective and accurate grouping than what can be achieved by just using one view of data. Multi-view clustering methods differ in the kind of modelling they use in order to fuse multiple views, by managing the synergies, complimentarities, and conflicts across data views, and arriving at a single clustering output across the multiple views in the dataset. This chapter provides a survey of a sample of multi-view clustering methods, with an emphasis on bringing out the wide diversity in solution formulations that have been considered. We pay specific attention to enable the reader understand the intuition behind each method ahead of describing the technical details of the method, to ensure that the survey is accessible to readers who may not be machine learning specialists. We also outline some popular datasets that have been used to empirically evaluate MVC methods
Domain-specific lexicon generation for emotion detection from text.
Full text linkEmotions play a key role in effective and successful human communication. Text is popularly used on the internet and social media websites to express and share emotions, feelings and sentiments. However useful applications and services built to understand emotions from text are limited in effectiveness due to reliance on general purpose emotion lexicons that have static vocabulary and sentiment lexicons that can only interpret emotions coarsely. Thus emotion detection from text calls for methods and knowledge resources that can deal with challenges such as dynamic and informal vocabulary, domain-level variations in emotional expressions and other linguistic nuances. In this thesis we demonstrate how labelled (e.g. blogs, news headlines) and weakly-labelled (e.g. tweets) emotional documents can be harnessed to learn word-emotion lexicons that can account for dynamic and domain-specific emotional vocabulary. We model the characteristics of realworld emotional documents to propose a generative mixture model, which iteratively estimates the language models that best describe the emotional documents using expectation maximization (EM). The proposed mixture model has the ability to model both emotionally charged words and emotion-neutral words. We then generate a word-emotion lexicon using the mixture model to quantify word-emotion associations in the form of a probability vectors. Secondly we introduce novel feature extraction methods to utilize the emotion rich knowledge being captured by our word-emotion lexicon. The extracted features are used to classify text into emotion classes using machine learning. Further we also propose hybrid text representations for emotion classification that use the knowledge of lexicon based features in conjunction with other representations such as n-grams, part-of-speech and sentiment information. Thirdly we propose two different methods which jointly use an emotion-labelled corpus of tweets and emotion-sentiment mapping proposed in psychology to learn word-level numerical quantification of sentiment strengths over a positive to negative spectrum. Finally we evaluate all the proposed methods in this thesis through a variety of emotion detection and sentiment analysis tasks on benchmark data sets covering domains from blogs to news articles to tweets and incident reports
Building Thermal Performance Varies During Extreme Heat within Cities
No full textabstract: This document has been superseded by our peer-reviewed publication:
Building Thermal Performance, Climate Change, and Urban Heat Vulnerability, Matthew Nahlik, Mikhail Chester, Stephanie Pincetl, David Eisenman, Deepak Sivaraman, and Paul English, 2017, ASCE Journal of Infrastructure Systems, 23(3), doi:10.1061/(ASCE)IS.1943-555X.0000349
The publication is available at: http://dx.doi.org/10.1061/(ASCE)IS.1943-555X.0000349
The leading source of weather-related deaths in the United States is heat, and future projections show that the frequency, duration, and intensity of heat events will increase in the Southwest. Presently, there is a dearth of knowledge about how infrastructure may perform during heat waves or could contribute to social vulnerability. To understand how buildings perform in heat and potentially stress people, indoor air temperature changes when air conditioning is inaccessible are modeled for building archetypes in Los Angeles, California, and Phoenix, Arizona, when air conditioning is inaccessible is estimated. An energy simulation model is used to estimate how quickly indoor air temperature changes when building archetypes are exposed to extreme heat. Building age and geometry (which together determine the building envelope material composition) are found to be the strongest indicators of thermal envelope performance. Older neighborhoods in Los Angeles and Phoenix (often more centrally located in the metropolitan areas) are found to contain the buildings whose interiors warm the fastest, raising particular concern because these regions are also forecast to experience temperature increases. To combat infrastructure vulnerability and provide heat refuge for residents, incentives should be adopted to strategically retrofit buildings where both socially vulnerable populations reside and increasing temperatures are forecast
Building and processing a dataset containing articles related to food adulteration
No full textThesis: M. Eng., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2015.This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.Cataloged from student-submitted PDF version of thesis.Includes bibliographical references (page 69).In this thesis, I explored the problem of building a dataset containing news articles related to adulteration, and curating this dataset in an automated fashion. In particular, we looked at food-adulterant co-existence detection, query reforumulation, and entity extraction and text deduplication. All proposed algorithms were implemented in Python, and performance was evaluated on multiple datasets. Methods described in this thesis can be generalized to other applications as well.by Deepak Narayanan.M. Eng
Digital front end for base-station RF
No full textThe digital front-end (DFE) is the most critical stage in a wireless base-station. The DFE along with the analog to digital converter (ADC) is responsible for bridging the analog RF and IF processing on one side and the digital baseband processing on the other side. The most important reason for replacing analog with digital signal processing is the ability to softly reconfigure the channels in the base station RF in real time, thus allowing for the implementation of various signal conditioning, compensation and mitigation channel non-linear responses. Once tested, these algorithms can be implemented on a proprietary CMOS vector processor and commercial FPGA hardware platforms. In this thesis, we attempt to minimize the design efforts and lower the cost involved in the usage of analog electronics by using sophisticated digital signal processing (DSP) for restoring and enhancing the quality of the wireless channels. This thesis presents a versatile Digital Front-End architecture, which has been simulated using MATLAB/Simulink. The architecture includes the design of robust Digital Up-Conversion (DUC) blocks in the transmit downlink and Digital Down-Conversion (DDC) blocks present in the receiver uplink paths in a wireless base station RF. Crest factor reduction (CFR) schemes help reduce the Peak to Average Power Ratio (PAPR)of the signal entering the base-station and have been implemented widely for code division multiple access (CDMA) and Long Term Evolution (LTE) systems, this is important because if the signal with the high PAPR is allowed to pass through the power amplifier(PA) it will result in the amplifier operating in its nonlinear region creating non-linear distortions in amplitude and phase, and the only other way to avoid this is to back off the signal to the linear region of the amplifier thus reducing its efficiency. The selection and design of the DUC and DDC filters has been compared and optimized to match to the spectral mask requirements mentioned in the 3GPP standards. Crest factor reduction has also been studied in detail and a computationally efficient algorithm for meeting the desired PAPR in accordance with the 3GPP standards will be presented. By using the CFR algorithm, the PAPR of the LTE signal was reduced from 10.8 dB to 7 dB and from 10.5 dB to 8 dB for a WCDMA signal. Finally, we implement Digital Predistortion (DPD) which is a method by which one first stimulates a non-linear power amplifier (PA) with baseband samples and then observes the result of that stimulus at its output. Without this process we will need to use a power amplifier with a higher input power rating which needs to be backed off to operate in its linear region thus reducing the efficiency of the PA used and increasing its cost. The process involves the use of a digital predistorter which creates an expanding nonlinearity which when used in cascade with the PA nullifies the compressing nonlinear characteristics of the PA thus enabling its use in its linear region up to its saturating point. A Look-Up Table (LUT) type Adaptive Digital Pre-Distortion (ADPD) is presented; here we developed an algorithm where the output signal of the PA is used as a reference signal. This reference signal is then used to update the coefficients of the LUT, so that the non-linear responses of the PA will not the affect the amplified signals. In addition, we investigated methods such as the nonlinear auto-regressive moving average (NARMA) and the memory polynomial models. In the latter, the predistorter parameters are calculated from the output signal obtained from the PA through the adaptive functions obtained using the memory polynomial. From these parameters, the predistorted signal is reconstructed and fed to the input of the PA. By using the DPD algorithm the nonlinear distortions of the PA came down by 60 dB when a WCDMA signal was used and by around 40 dB when LTE signal was used. As the PA is the heart of the base-station RF, we show that the main function of the DFE is to ensure a PA linearized output with a high efficiency.M.S.Includes bibliographical referencesby Deepak Gop
- …
