“I had the privilege of studying with Prem. He has a fantastic ability of simplifying complex problems. Prem has a very high intellect combined with ability of coming out with practical solutions. Deceptively simple, whacky sense of humour and great human being.”
Prem Natarajan, PhD
Marina del Rey, California, United States
8K followers
500+ connections
Activity
-
Congratulations to our first cohort of PhD and faculty research award recipients at the Center for AI and Responsible Financial Innovation alongside…
Congratulations to our first cohort of PhD and faculty research award recipients at the Center for AI and Responsible Financial Innovation alongside…
Shared by Prem Natarajan, PhD
-
Thanks, H2O.ai! It's an honor to be recognized alongside this accomplished group of leaders -- and a testament to the passion and talent of the…
Thanks, H2O.ai! It's an honor to be recognized alongside this accomplished group of leaders -- and a testament to the passion and talent of the…
Shared by Prem Natarajan, PhD
-
H2O.ai Celebrates Global Leaders Driving AI in First-Ever AI 100 List At H2O.ai, we’re celebrating leaders from both the private and public sectors…
H2O.ai Celebrates Global Leaders Driving AI in First-Ever AI 100 List At H2O.ai, we’re celebrating leaders from both the private and public sectors…
Liked by Prem Natarajan, PhD
Publications
-
A Constrained Optimization Approach to Combining Multiple Non-Local Means Denoising Estimates
Signal Processing
There is an ongoing need to develop image denoising approaches that suppress noise while maintaining edge information. The non-local means (NLM) algorithm, a widely used patch-based method, is a highly effective edge-preserving technique but is sensitive to parameter tuning. We use a variational approach to combine multiple NLM estimates, seeking a solution that balances positivity constraints and gradient penalties against Stein's Unbiased Risk Estimate (SURE). This method greatly reduces…
There is an ongoing need to develop image denoising approaches that suppress noise while maintaining edge information. The non-local means (NLM) algorithm, a widely used patch-based method, is a highly effective edge-preserving technique but is sensitive to parameter tuning. We use a variational approach to combine multiple NLM estimates, seeking a solution that balances positivity constraints and gradient penalties against Stein's Unbiased Risk Estimate (SURE). This method greatly reduces parameter sensitivity and improves denoising performance vs. other NLM variants.
Other authorsSee publication -
Robust named entity detection from optical character recognition output
International Journal on Document Analysis and Recognition (IJDAR) June 2011, Volume 14, Issue 2, pp 189-200
In this paper, we focus on information extraction from optical character recognition (OCR) output. Since the content from OCR inherently has many errors, we present robust algorithms for information extraction from OCR lattices instead of merely looking them up in the top-choice (1-best) OCR output. Specifically, we address the challenge of named entity detection in noisy OCR output and show that searching for named entities in the recognition lattice significantly improves detection accuracy…
In this paper, we focus on information extraction from optical character recognition (OCR) output. Since the content from OCR inherently has many errors, we present robust algorithms for information extraction from OCR lattices instead of merely looking them up in the top-choice (1-best) OCR output. Specifically, we address the challenge of named entity detection in noisy OCR output and show that searching for named entities in the recognition lattice significantly improves detection accuracy over 1-best search. While lattice-based named entity (NE) detection improves NE recall from OCR output, there are two problems with this approach: (1) the number of false alarms can be prohibitive for certain applications and (2) lattice-based search is computationally more expensive than 1-best NE lookup. To mitigate the above challenges, we present techniques for reducing false alarms using confidence measures and for reducing the amount of computation involved in performing the NE search. Furthermore, to demonstrate that our techniques are applicable across multiple domains and languages, we experiment with optical character recognition systems for videotext in English and scanned handwritten text in Arabic.
Other authorsSee publication -
Stochastic Segment Modeling for Offline Handwriting Recognition
Document Analysis and Recognition, 2009. ICDAR '09. 10th International Conference on
In this paper, we present a novel approach for incorporating structural information into the hidden Markov modeling (HMM) framework for offline handwriting recognition. Traditionally, structural features have been used in recognition approaches that rely on accurate segmentation of words into smaller units (sub-words or characters). However, such segmentation based approaches do not perform well on real-world handwritten images, because breaks and merges in glyphs typically create new connected…
In this paper, we present a novel approach for incorporating structural information into the hidden Markov modeling (HMM) framework for offline handwriting recognition. Traditionally, structural features have been used in recognition approaches that rely on accurate segmentation of words into smaller units (sub-words or characters). However, such segmentation based approaches do not perform well on real-world handwritten images, because breaks and merges in glyphs typically create new connected components that are not observed in the training data. To mitigate the problem of having to derive accurate segmentation from connected components, we present a novel framework where the HMM based recognition system trained on shorter-span features is used to generate the 2D character images (the ldquostochastic segmentsrdquo), and then another classifier that uses structural features extracted from the stochastic character segments generates a new set of scores. Finally, the scores from the HMM system and from structural matching are used in combination to generate a hypothesis that is better than the results from either the HMM or from structural matching alone. We demonstrate the efficacy of our approach by reporting experimental results on a large corpus of handwritten Arabic documents.
Other authorsSee publication -
A Wearable Headset Speech-to-Speech Translation System
Proc. ACL 2008 Workshop on Mobile Language Processing
We present a wearable, headset integrated eyes- and hands-free speech-to-speech (S2S) translation system. It employs an n-gram speech recognition engine, a rudimentary phrase-based translator for translating recognized text, and a rudimentary text-to speech (TTS) synthesis engine for playing back the English translation. [Pretty good for 2008, if I do say so myself.]
Other authorsSee publication -
Character-Stroke Detection for Text Localization and Extraction
9th International Conference on Document Analysis and Recognition (ICDAR 2007)
We present a new approach for analysis of images for text localization and extraction. Our approach puts very few constraints on the font, size and color of text and is capable of handling both scene text and artificial text well. In this paper, we exploit two well-known features of text: approximately constant stroke width and local contrast, and develop a fast, simple, and effective algorithm to detect character strokes. We also show how these can be used for accurate extraction and motivate…
We present a new approach for analysis of images for text localization and extraction. Our approach puts very few constraints on the font, size and color of text and is capable of handling both scene text and artificial text well. In this paper, we exploit two well-known features of text: approximately constant stroke width and local contrast, and develop a fast, simple, and effective algorithm to detect character strokes. We also show how these can be used for accurate extraction and motivate some advantages of using this approach for text localization over other color-space segmentation based approaches. We analyze the performance of our stroke detection algorithm on images col- lected for the robust-reading competitions at ICDAR 2003.
Other authorsSee publication -
Optimal Estimation of Rejection Thresholds for Topic Spotting
Acoustics, Speech and Signal Processing, 2007. ICASSP 2007. IEEE International Conference on
In many applications of topic spotting technology, especially those that require a human review of in-topic documents, a low false alarm rate is a key requirement. Topic spotting techniques typically include a rejection scheme to filter out off-topic documents. In this paper we present a robust methodology for rejecting off-topic messages that, in addition to modeling the topics of interest, uses a so-called alternate model for topics that are not included in the set of topics of interest…
In many applications of topic spotting technology, especially those that require a human review of in-topic documents, a low false alarm rate is a key requirement. Topic spotting techniques typically include a rejection scheme to filter out off-topic documents. In this paper we present a robust methodology for rejecting off-topic messages that, in addition to modeling the topics of interest, uses a so-called alternate model for topics that are not included in the set of topics of interest. Specifically, we introduce two novel techniques for estimating topic-specific rejection thresholds - a parametric technique that can be viewed as transformation of topic-independent thresholds, and a nonparametric technique based on constrained optimization of false rejections subject to a pre-specified number of false acceptances. Our experiments on newsgroup messages demonstrate that when adequate training data is available topic-specific threshold estimation techniques can outperform topic-independent thresholds in terms of the ROC curve.
Other authorsSee publication
Patents
-
Home Call Router
Issued US 8340262
Disclosed is a method and system for routing telephone calls within a household. In the disclosed home call routing system, a head of the household or other person with administrative authority within the home can control the routing of telephone calls by establishing and modifying call system parameters such as call priorities, traffic times, caller identities, routing rules, etc. through a home computer.
-
Multiframe Videotext Recognition
Issued US 8,290,273
Multi-frame persistence of videotext is exploited to mitigate challenges posed by varying characteristics of videotext across frame instances to improve OCR techniques. In some examples, each frame of video is processed to form multiple binary images, and one or more text hypotheses is formed from each binary image. In some examples, one or more combined images are formed from multiple frames processed to form a binary image and a corresponding text hypothesis. The text hypotheses are combined…
Multi-frame persistence of videotext is exploited to mitigate challenges posed by varying characteristics of videotext across frame instances to improve OCR techniques. In some examples, each frame of video is processed to form multiple binary images, and one or more text hypotheses is formed from each binary image. In some examples, one or more combined images are formed from multiple frames processed to form a binary image and a corresponding text hypothesis. The text hypotheses are combined to yield an overall text recognition output.
Other inventors -
-
Method and apparatus for training an automated speech recognition-based system
Issued US 7346507
A method and apparatus for building a training set for an automated speech recognition-based system, which determines the statistically optimal number of frequently requested responses to automate in order to achieve a desired automation rate. The invention may be used to select the appropriate tokens and responses to train the system and to achieve a desired "phrase coverage" for all of the many different ways human beings may phrase a request that calls for one of a plurality of…
A method and apparatus for building a training set for an automated speech recognition-based system, which determines the statistically optimal number of frequently requested responses to automate in order to achieve a desired automation rate. The invention may be used to select the appropriate tokens and responses to train the system and to achieve a desired "phrase coverage" for all of the many different ways human beings may phrase a request that calls for one of a plurality of frequently-requested responses. The invention also determines the statistically optimal number of tokens (spoken requests) required to train a speech recognition-based system to achieve the desired phrase coverage and optimal allocation of tokens over the set of responses that are to be automated.
Other inventors -
-
Unsupervised Training in Natural Language Call Routing
Issued US 7092888
A method of training a natural language call routing system using an unsupervised trainer is provided. The unsupervised trainer is adapted to tune performance of the call routing system on the basis of feedback and new topic information. The method of training comprises: storing audio data from an incoming call as well as associated unique identifier information for the incoming call; applying a highly accurate speech recognizer to the audio data from the waveform database to produce a text…
A method of training a natural language call routing system using an unsupervised trainer is provided. The unsupervised trainer is adapted to tune performance of the call routing system on the basis of feedback and new topic information. The method of training comprises: storing audio data from an incoming call as well as associated unique identifier information for the incoming call; applying a highly accurate speech recognizer to the audio data from the waveform database to produce a text transcription of the stored audio for the call; forwarding outputs of the second speech recognizer to a training database, the training database being adapted to store text transcripts from the second recognizer with respective unique call identifiers as well as topic data; for a call routed by the call router to an agent: entering a call topic determined by the agent into a form; and supplying the call topic information from the form to the training database together with the associated unique call identifier; and for a call routed to automated fulfillment: querying the caller regarding the true topic of the call; and adding this topic information, together with the associated unique call identifier, to the training database; and performing topic identification model training and statistical grammar model training on the basis of the topic information and transcription information stored in the training database.
Other inventorsSee patent
Recommendations received
2 people have recommended Prem
Join now to viewMore activity by Prem
-
Thank you, Randy Bean, for the great reporting. Capital One has a rich legacy of being powered by data and analytics, and that expertise combined…
Thank you, Randy Bean, for the great reporting. Capital One has a rich legacy of being powered by data and analytics, and that expertise combined…
Shared by Prem Natarajan, PhD
-
I recently returned from an amazing trip to Bengaluru, India where I, along with Prem Natarajan, PhD, Milind Naphade and Aparna Sinha had the…
I recently returned from an amazing trip to Bengaluru, India where I, along with Prem Natarajan, PhD, Milind Naphade and Aparna Sinha had the…
Liked by Prem Natarajan, PhD
-
We are thrilled to announce our inaugural investment from IA Global Fund. Congratulations Delian Coroama and the rest of the Recomaze AI team and…
We are thrilled to announce our inaugural investment from IA Global Fund. Congratulations Delian Coroama and the rest of the Recomaze AI team and…
Liked by Prem Natarajan, PhD
-
🚀 Announcing the 2024 MIT Digital Technology and Strategy Conference! 📅 September 17-18, 2024 Join us to explore the forefront of digital…
🚀 Announcing the 2024 MIT Digital Technology and Strategy Conference! 📅 September 17-18, 2024 Join us to explore the forefront of digital…
Liked by Prem Natarajan, PhD
-
I am thrilled to re-join the Partnership on AI (PAI) Board. PAI's mission is more relevant today than ever. Now is a critical time to work across…
I am thrilled to re-join the Partnership on AI (PAI) Board. PAI's mission is more relevant today than ever. Now is a critical time to work across…
Shared by Prem Natarajan, PhD
-
300+ people at Tau Ventures 6th Annual Day! Many thanks to all those who joined and to our entire global village. Collaboration and cooperation is…
300+ people at Tau Ventures 6th Annual Day! Many thanks to all those who joined and to our entire global village. Collaboration and cooperation is…
Liked by Prem Natarajan, PhD
-
Bingjie Tang will present AutoMate at #RSS2024. AutoMate provides the first simulation-based framework for learning specialist and generalist…
Bingjie Tang will present AutoMate at #RSS2024. AutoMate provides the first simulation-based framework for learning specialist and generalist…
Liked by Prem Natarajan, PhD
Other similar profiles
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More