Buy Ensemble Feature Selection for Multi-Stream Automatic Speech Recognition
Book 1
Book 2
Book 3
Book 1
Book 2
Book 3
Book 1
Book 2
Book 3
Book 1
Book 2
Book 3
Home > Science, Technology & Agriculture > Technology: general issues > Ensemble Feature Selection for Multi-Stream Automatic Speech Recognition
Ensemble Feature Selection for Multi-Stream Automatic Speech Recognition

Ensemble Feature Selection for Multi-Stream Automatic Speech Recognition


     0     
5
4
3
2
1



Out of Stock


Notify me when this book is in stock
X
About the Book

Multi-stream automatic speech recognition (ASR) systems consisting of an ensemble of classifiers working together, each with its own feature vector, are popular in the research literature. Published work on feature selection for such systems has dealt with indivisible blocks of features. I break from this tradition by investigating feature selection at the level of individual features. I use the OGI ISOLET and Numbers speech corpora, including noisy versions I created using a variety of noises and signal-to-noise ratios. I have made these noisy versions available for use by other researchers, along with my ASR and feature selection scripts. I start with the random subspace method of ensemble feature selection, in which each feature vector is simply chosen randomly from the feature pool. Using ISOLET, I obtain performance improvements over baseline in almost every case where there is a statistically significant performance difference, but there are many cases with no such difference. I then try hill-climbing, a wrapper approach that changes a single feature at a time when the change improves a performance score. With ISOLET, hill-climbing gives performance improvements in most cases for noisy data, but no improvement for clean data. I then move to Numbers, for which much more data is available to guide hill-climbing. When using either the clean or noisy Numbers data, hill-climbing gives performance improvements over multi-stream baselines in almost all cases, although it does not improve over the best single-stream baseline. For noisy data, these performance improvements are present even for noise types that were not seen during the hill-climbing process. In mismatched condition tests involving mismatch between clean and noisy data, hill-climbing outperforms all baselines when Opitz's scoring formula is used. I find that this scoring formula, which blends single-classifier accuracy and ensemble diversity, works better for me than ensemble accuracy as a performance score for guiding hill-climbing.


Best Sellers


Product Details
  • ISBN-13: 9781243998989
  • Publisher: Proquest, Umi Dissertation Publishing
  • Publisher Imprint: Proquest, Umi Dissertation Publishing
  • Height: 246 mm
  • Weight: 240 gr
  • ISBN-10: 1243998989
  • Publisher Date: 01 Sep 2011
  • Binding: Paperback
  • Spine Width: 7 mm
  • Width: 189 mm


Similar Products

Add Photo
Add Photo

Customer Reviews

REVIEWS      0     
Click Here To Be The First to Review this Product
Ensemble Feature Selection for Multi-Stream Automatic Speech Recognition
Proquest, Umi Dissertation Publishing -
Ensemble Feature Selection for Multi-Stream Automatic Speech Recognition
Writing guidlines
We want to publish your review, so please:
  • keep your review on the product. Review's that defame author's character will be rejected.
  • Keep your review focused on the product.
  • Avoid writing about customer service. contact us instead if you have issue requiring immediate attention.
  • Refrain from mentioning competitors or the specific price you paid for the product.
  • Do not include any personally identifiable information, such as full names.

Ensemble Feature Selection for Multi-Stream Automatic Speech Recognition

Required fields are marked with *

Review Title*
Review
    Add Photo Add up to 6 photos
    Would you recommend this product to a friend?
    Tag this Book Read more
    Does your review contain spoilers?
    What type of reader best describes you?
    I agree to the terms & conditions
    You may receive emails regarding this submission. Any emails will include the ability to opt-out of future communications.

    CUSTOMER RATINGS AND REVIEWS AND QUESTIONS AND ANSWERS TERMS OF USE

    These Terms of Use govern your conduct associated with the Customer Ratings and Reviews and/or Questions and Answers service offered by Bookswagon (the "CRR Service").


    By submitting any content to Bookswagon, you guarantee that:
    • You are the sole author and owner of the intellectual property rights in the content;
    • All "moral rights" that you may have in such content have been voluntarily waived by you;
    • All content that you post is accurate;
    • You are at least 13 years old;
    • Use of the content you supply does not violate these Terms of Use and will not cause injury to any person or entity.
    You further agree that you may not submit any content:
    • That is known by you to be false, inaccurate or misleading;
    • That infringes any third party's copyright, patent, trademark, trade secret or other proprietary rights or rights of publicity or privacy;
    • That violates any law, statute, ordinance or regulation (including, but not limited to, those governing, consumer protection, unfair competition, anti-discrimination or false advertising);
    • That is, or may reasonably be considered to be, defamatory, libelous, hateful, racially or religiously biased or offensive, unlawfully threatening or unlawfully harassing to any individual, partnership or corporation;
    • For which you were compensated or granted any consideration by any unapproved third party;
    • That includes any information that references other websites, addresses, email addresses, contact information or phone numbers;
    • That contains any computer viruses, worms or other potentially damaging computer programs or files.
    You agree to indemnify and hold Bookswagon (and its officers, directors, agents, subsidiaries, joint ventures, employees and third-party service providers, including but not limited to Bazaarvoice, Inc.), harmless from all claims, demands, and damages (actual and consequential) of every kind and nature, known and unknown including reasonable attorneys' fees, arising out of a breach of your representations and warranties set forth above, or your violation of any law or the rights of a third party.


    For any content that you submit, you grant Bookswagon a perpetual, irrevocable, royalty-free, transferable right and license to use, copy, modify, delete in its entirety, adapt, publish, translate, create derivative works from and/or sell, transfer, and/or distribute such content and/or incorporate such content into any form, medium or technology throughout the world without compensation to you. Additionally,  Bookswagon may transfer or share any personal information that you submit with its third-party service providers, including but not limited to Bazaarvoice, Inc. in accordance with  Privacy Policy


    All content that you submit may be used at Bookswagon's sole discretion. Bookswagon reserves the right to change, condense, withhold publication, remove or delete any content on Bookswagon's website that Bookswagon deems, in its sole discretion, to violate the content guidelines or any other provision of these Terms of Use.  Bookswagon does not guarantee that you will have any recourse through Bookswagon to edit or delete any content you have submitted. Ratings and written comments are generally posted within two to four business days. However, Bookswagon reserves the right to remove or to refuse to post any submission to the extent authorized by law. You acknowledge that you, not Bookswagon, are responsible for the contents of your submission. None of the content that you submit shall be subject to any obligation of confidence on the part of Bookswagon, its agents, subsidiaries, affiliates, partners or third party service providers (including but not limited to Bazaarvoice, Inc.)and their respective directors, officers and employees.

    Accept

    Fresh on the Shelf


    Inspired by your browsing history


    Your review has been submitted!

    You've already reviewed this product!