69 Hits in 1.3 sec

Recursive Neural Language Architecture for Tag Prediction [article]

Saurabh Kataria
2016 arXiv   pre-print
We consider the problem of learning distributed representations for tags from their associated content for the task of tag recommendation. Considering tagging information is usually very sparse, effective learning from content and tag association is very crucial and challenging task. Recently, various neural representation learning models such as WSABIE and its variants show promising performance, mainly due to compact feature representations learned in a semantic space. However, their capacity
more » ... is limited by a linear compositional approach for representing tags as sum of equal parts and hurt their performance. In this work, we propose a neural feedback relevance model for learning tag representations with weighted feature representations. Our experiments on two widely used datasets show significant improvement for quality of recommendations over various baselines.
arXiv:1603.07646v1 fatcat:m55vhpse4zb57g564wrdawxgoy

Multitask Learning for Sequence Labeling Tasks [article]

Arvind Agarwal, Saurabh Kataria
2016 arXiv   pre-print
In this paper, we present a learning method for sequence labeling tasks in which each example sequence has multiple label sequences. Our method learns multiple models, one model for each label sequence. Each model computes the joint probability of all label sequences given the example sequence. Although each model considers all label sequences, its primary focus is only one label sequence, and therefore, each model becomes a task-specific model, for the task belonging to that primary label.
more » ... multiple models are learned simultaneously by facilitating the learning transfer among models through explicit parameter sharing. We experiment the proposed method on two applications and show that our method significantly outperforms the state-of-the-art method.
arXiv:1404.6580v2 fatcat:4gl4mnrtrfarxmi5gtd22f3ovy

Long-Term Memory Networks for Question Answering [article]

Fenglong Ma, Radha Chitta, Saurabh Kataria, Jing Zhou, Palghat Ramesh, Tong Sun, Jing Gao
2017 arXiv   pre-print
Question answering is an important and difficult task in the natural language processing domain, because many basic natural language processing tasks can be cast into a question answering task. Several deep neural network architectures have been developed recently, which employ memory and inference components to memorize and reason over text information, and generate answers to questions. However, a major drawback of many such models is that they are capable of only generating single-word
more » ... s. In addition, they require large amount of training data to generate accurate answers. In this paper, we introduce the Long-Term Memory Network (LTMN), which incorporates both an external memory module and a Long Short-Term Memory (LSTM) module to comprehend the input data and generate multi-word answers. The LTMN model can be trained end-to-end using back-propagation and requires minimal supervision. We test our model on two synthetic data sets (based on Facebook's bAbI data set) and the real-world Stanford question answering data set, and show that it can achieve state-of-the-art performance.
arXiv:1707.01961v1 fatcat:6jiqnvogojd2jabgfijy4js7fe

Low-Resource Domain Adaptation for Speaker Recognition Using Cycle-GANs [article]

Phani Sankar Nidadavolu, Saurabh Kataria, Jesús Villalba, Najim Dehak
2019 arXiv   pre-print
Current speaker recognition technology provides great performance with the x-vector approach. However, performance decreases when the evaluation domain is different from the training domain, an issue usually addressed with domain adaptation approaches. Recently, unsupervised domain adaptation using cycle-consistent Generative Adversarial Netorks (CycleGAN) has received a lot of attention. CycleGAN learn mappings between features of two domains given non-parallel data. We investigate their
more » ... iveness in low resource scenario i.e. when limited amount of target domain data is available for adaptation, a case unexplored in previous works. We experiment with two adaptation tasks: microphone to telephone and a novel reverberant to clean adaptation with the end goal of improving speaker recognition performance. Number of speakers present in source and target domains are 7000 and 191 respectively. By adding noise to the target domain during CycleGAN training, we were able to achieve better performance compared to the adaptation system whose CycleGAN was trained on a larger target data. On reverberant to clean adaptation task, our models improved EER by 18.3% relative on VOiCES dataset compared to a system trained on clean data. They also slightly improved over the state-of-the-art Weighted Prediction Error (WPE) de-reverberation algorithm.
arXiv:1910.11909v1 fatcat:wcfzlbtzvjcyxcnstc32757qpu

VAST: The Virtual Acoustic Space Traveler Dataset [chapter]

Clément Gaultier, Saurabh Kataria, Antoine Deleforge
2017 Lecture Notes in Computer Science  
This paper introduces a new paradigm for sound source localization referred to as virtual acoustic space traveling (VAST) and presents a first dataset designed for this purpose. Existing sound source localization methods are either based on an approximate physical model (physics-driven) or on a specific-purpose calibration set (data-driven). With VAST, the idea is to learn a mapping from audio features to desired audio properties using a massive dataset of simulated room impulse responses. This
more » ... virtual dataset is designed to be maximally representative of the potential audio scenes that the considered system may be evolving in, while remaining reasonably compact. We show that virtually-learned mappings on this dataset generalize to real data, overcoming some intrinsic limitations of traditional binaural sound localization methods based on time differences of arrival.
doi:10.1007/978-3-319-53547-0_7 fatcat:gzlxjunjbrdazpd4q3ue67ml7q

Unsupervised Feature Enhancement for speaker verification [article]

Phani Sankar Nidadavolu, Saurabh Kataria, Jesús Villalba, Paola García-Perera, Najim Dehak
2020 arXiv   pre-print
The task of making speaker verification systems robust to adverse scenarios remain a challenging and an active area of research. We developed an unsupervised feature enhancement approach in log-filter bank domain with the end goal of improving speaker verification performance. We experimented with using both real speech recorded in adverse environments and degraded speech obtained by simulation to train the enhancement systems. The effectiveness of the approach was shown by testing on several
more » ... al, simulated noisy, and reverberant test sets. The approach yielded significant improvements on both real and simulated sets when data augmentation was not used in speaker verification pipeline or augmentation was used only during x-vector training. When data augmentation was used for x-vector and PLDA training, our enhancement approach yielded slight improvements.
arXiv:1910.11915v2 fatcat:rcih4ckssbhqriuoedx6l6yeqy

Dentigerous Cyst: A Review of Literature

Saurabh Bhardwaj, Mehak Anand, Gazalla Altaf, Sakshi Kataria
2019 International Healthcare Research Journal  
REFRENCES Figure 1 . 1 Microphotograph : Dentigerous Cyst, Odontogenic Keratocyst, Enucleation, Marsupialization SAURABH BHARDWAJ 1 , MEHAK ANAND* 2 , GAZALLA ALTAF 3 , SAKSHI KATARIA 4 © Saurabh Bhardwaj  ...  Journal of Dentigerous Cyst: A Review of Literature Kataria S et al. Source of support: Nil, Conflict of interest: None declared  ... 
doi:10.26440/ihrj/0302.05.521077 fatcat:cnnnb6bb6bfyfios62rqj3er4q

Automatic Identification and Data Extraction from 2-Dimensional Plots in Digital Documents [article]

William Brouwer, Saurabh Kataria, Sujatha Das, Prasenjit Mitra, C. L. Giles
2008 arXiv   pre-print
Most search engines index the textual content of documents in digital libraries. However, scholarly articles frequently report important findings in figures for visual impact and the contents of these figures are not indexed. These contents are often invaluable to the researcher in various fields, for the purposes of direct comparison with their own work. Therefore, searching for figures and extracting figure data are important problems. To the best of our knowledge, there exists no tool to
more » ... matically extract data from figures in digital documents. If we can extract data from these images automatically and store them in a database, an end-user can query and combine data from multiple digital documents simultaneously and efficiently. We propose a framework based on image analysis and machine learning to extract information from 2-D plot images and store them in a database. The proposed algorithm identifies a 2-D plot and extracts the axis labels, legend and the data points from the 2-D plot. We also segregate overlapping shapes that correspond to different data points. We demonstrate performance of individual algorithms, using a combination of generated and real-life images.
arXiv:0809.1802v1 fatcat:dxeuw7aukbb3bn5xq523v6nfbe

Trigeminal Neuralgia Induced Headache: A Case Report and Literature Review

Saurabh Kataria, Zahoor Ahmed, Unaiza Ali, Sarfaraz Ahmad, Anum Awais
2020 Cureus  
DOI 10.7759/cureus.9226 2 of 3 Kataria et al. Cureus 12(7): e9226. DOI 10.7759/cureus.9226 3 of 3  ...  TABLE 1 : 1 Results of hematological examination and metabolic panel.WBC, white blood cells; RBC, red blood cells; AST, aspartate aminotransferase; ALT, alanine aminotransferase Kataria et al.  ... 
doi:10.7759/cureus.9226 pmid:32821575 pmcid:PMC7430534 fatcat:l4tcy34uorb7tgojjec25f56li

Single Channel Far Field Feature Enhancement For Speaker Verification In The Wild [article]

Phani Sankar Nidadavolu, Saurabh Kataria, Paola García-Perera, Jesús Villalba, Najim Dehak
2020 arXiv   pre-print
We investigated an enhancement and a domain adaptation approach to make speaker verification systems robust to perturbations of far-field speech. In the enhancement approach, using paired (parallel) reverberant-clean speech, we trained a supervised Generative Adversarial Network (GAN) along with a feature mapping loss. For the domain adaptation approach, we trained a Cycle Consistent Generative Adversarial Network (CycleGAN), which maps features from far-field domain to the speaker embedding
more » ... ining domain. This was trained on unpaired data in an unsupervised manner. Both networks, termed Supervised Enhancement Network (SEN) and Domain Adaptation Network (DAN) respectively, were trained with multi-task objectives in (filter-bank) feature domain. On a simulated test setup, we first note the benefit of using feature mapping (FM) loss along with adversarial loss in SEN. Then, we tested both supervised and unsupervised approaches on several real noisy datasets. We observed relative improvements ranging from 2% to 31% in terms of DCF. Using three training schemes, we also establish the effectiveness of the novel DAN approach.
arXiv:2005.08331v1 fatcat:blfs4zinozbkrcgisd3tgj5goa

Primary Periodic Paralyses: A Review of Etiologies and Their Pathogeneses

Umar Farooque, Asfand Yar Cheema, Ranjeet Kumar, Gagandeep Saini, Saurabh Kataria
2020 Cureus  
Periodic paralyses are a group of disorders characterized by episodes of muscle paralyses. They are mainly divided as primary (hereditary) and secondary (acquired) periodic paralyses. Primary periodic paralyses occur as a result of mutations in genes encoding subunits of muscle membrane channel proteins such as sodium, calcium, and potassium channels, resulting in impairment of their properties. Primary periodic paralyses are further classified on the basis of affected ion channels and other
more » ... ociated complications. Some of these periodic paralyses are hyperkalemic periodic paralysis (Na-channel mutation), hypokalemic periodic paralysis (Na- or Ca-channel mutation), Andersen's syndrome (K-channel mutation), etc.
doi:10.7759/cureus.10112 pmid:33005530 pmcid:PMC7523540 fatcat:4eawmbekkjcjpgp42jkz4xybam

Coronavirus Disease 2019-Related Acute Ischemic Stroke: A Case Report

Umar Farooque, Sohaib Shabih, Sundas Karimi, Ashok Kumar Lohano, Saurabh Kataria
2020 Cureus  
Coronavirus disease 2019 (COVID-19) is an active worldwide pandemic with diverse presentations and complications. Most patients present with constitutional and respiratory symptoms. Acute ischemic stroke remains a medical emergency even during the COVID-19 pandemic. Here we present a case of a patient with COVID-19 who presented with acute ischemic stroke in the absence of common risk factors for cerebrovascular accidents. A 70-year-old male patient, with no prior comorbidities, presented to
more » ... emergency department (ED) with fever, cough, and shortness of breath for four days, and altered level of consciousness and right-sided weakness with the sensory loss for one day. On examination, the patient had a score of 8/15 on the Glasgow coma scale (GCS). There was a right-sided sensory loss and weakness in both upper and lower limbs with a positive Babinski's sign. The pulmonary examination was remarkable for bilateral crepitation. On blood workup, there was leukocytosis and raised c-reactive protein (CRP). D-dimer, ferritin, thyroid-stimulating hormone (TSH), vitamin B12, and hypercoagulability workup were normal. Transthoracic echocardiography was also normal. COVID-19 polymerase chain reaction (PCR) detected the virus. Chest x-ray showed infiltrations in the left middle and both lower zones of the lungs in the peripheral distribution. Computed tomography (CT) scan of the chest showed peripheral and mid to basal predominant multilobar ground-glass opacities. CT scan of the head showed a large hypodense area, with a loss of gray and white matter differentiation, in the left middle cerebral artery territory. Magnetic resonance imaging (MRI) of the head showed abnormal signal intensity area in the left parietal region. It appeared isointense on T1 image and hyperintense on T2 image. It also showed diffusion restriction on the diffusion-weighted 1 (DW1) image with corresponding low signals on the apparent diffusion coefficient (ADC) map. These findings were consistent with left middle cerebral artery territory infarct due to COVID-19. The patient was intubated in the ED. He was deemed unfit for thrombolysis and started on aspirin, anti-coagulation, and other supportive measures. Patients with COVID-19 should be evaluated early for neurological signs. Timely workup and interventions should be performed in any patient suspected of having a stroke to reduce morbidity and mortality.
doi:10.7759/cureus.10310 pmid:33052272 pmcid:PMC7544605 fatcat:cgy63trkpzeavaiontvwfc4a5m


Lei Li, Wei Peng, Saurabh Kataria, Tong Sun, Tao Li
2013 Proceedings of the 22nd ACM international conference on Conference on information & knowledge management - CIKM '13  
In this paper, we propose a framework of recommending users and communities in social media. Given a user's profile, our framework is capable of recommending influential users and topic-cohesive interactive communities that are most relevant to the given user. In our framework, we present a generative topic model to discover user-oriented and community-oriented topics simultaneously, which enables us to capture the exact topic interests of users, as well as the focuses of communities. Extensive
more » ... evaluation on a data set obtained from Twitter has demonstrated the effectiveness of our proposed framework compared with other probabilistic topic model based recommendation methods.
doi:10.1145/2505515.2505645 dblp:conf/cikm/LiPKSL13 fatcat:4ei2jtiy3bh3xkjqf4mnt4f7qa

Perceptual Loss based Speech Denoising with an ensemble of Audio Pattern Recognition and Self-Supervised Models [article]

Saurabh Kataria, Jesús Villalba, Najim Dehak
2020 arXiv   pre-print
Deep learning based speech denoising still suffers from the challenge of improving perceptual quality of enhanced signals. We introduce a generalized framework called Perceptual Ensemble Regularization Loss (PERL) built on the idea of perceptual losses. Perceptual loss discourages distortion to certain speech properties and we analyze it using six large-scale pre-trained models: speaker classification, acoustic model, speaker embedding, emotion classification, and two self-supervised speech
more » ... ders (PASE+, wav2vec 2.0). We first build a strong baseline (w/o PERL) using Conformer Transformer Networks on the popular enhancement benchmark called VCTK-DEMAND. Using auxiliary models one at a time, we find acoustic event and self-supervised model PASE+ to be most effective. Our best model (PERL-AE) only uses acoustic event model (utilizing AudioSet) to outperform state-of-the-art methods on major perceptual metrics. To explore if denoising can leverage full framework, we use all networks but find that our seven-loss formulation suffers from the challenges of Multi-Task Learning. Finally, we report a critical observation that state-of-the-art Multi-Task weight learning methods cannot outperform hand tuning, perhaps due to challenges of domain mismatch and weak complementarity of losses.
arXiv:2010.11860v1 fatcat:ofm2by6aynbixakhtck2cbpfni

A Case Report on Charcot-Marie-Tooth Disease with a Novel Periaxin Gene Mutation

Sorabh Datta, Saurabh Kataria, Raghav Govindarajan
2019 Cureus  
Charcot-Marie-Tooth (CMT) disease is one of the most common primary hereditary neuropathies causing peripheral neuropathies. More than 60 different gene mutations are causing this disease. The PRX gene codes for Periaxin proteins that are expressed by Schwann cells and are necessary for the formation and maintenance of myelination of peripheral nerves. Dejerine-Sottas neuropathy and Charcot-Marie-Tooth type 4F (CMT4F) are the two different clinical phenotypes observed in association with PRX
more » ... e mutation. This article describes a case of an elderly male with a novel mutation involving the PRX gene.
doi:10.7759/cureus.5111 pmid:31523542 pmcid:PMC6741374 fatcat:mw2vw2pplfb5zjy44purxd4d5y
« Previous Showing results 1 — 15 out of 69 results