• Documents
  • Authors
  • Tables
  • Log in
  • Sign up
  • MetaCart
  • DMCA
  • Donate

CiteSeerX logo

Tools

Sorted by:
Try your query at:
Semantic Scholar Scholar Academic
Google Bing DBLP
Results 1 - 10 of 11,247
Next 10 →

Table 5: Network Maximum Training Error (Scaled Training Data) accurate representation of the space.

in Optimization of Mixed Discrete/Continuous
by Design Variable Systems, R. S. Sellar, S. M. Batill, J. E. Renaud 1994
Cited by 7

Table 2: FIR Coe cient CSD Representation

in List of Figures
by Figure Block, Nader Bagherzadeh, Fadi J. Kurdahi, Hung-kang Liu, Farzad Etemadi, Guang-ming Lu
"... In PAGE 16: ... Simulation results showed that the two apos;s complement representation with a total number of 12 bits and one integer bit accurately quantizes the model. FIR lter taps were converted to their CSD representation using 1% and 5% relative errors and the results are shown in Table2 . The underlined digits will be ignored since they represent a shift of more than the bitwidth of 12.... ..."

Table 2. The representation of formality distinctions in the pronoun database

in Developing a database of personal and demonstrative pronoun paradigms: Conceptual and technical challenges
by Heather Bliss Elizabeth, Elizabeth Ritter
"... In PAGE 3: ... Less formal distinctions also proceed from 0, beginning with Neg1 (Negative 1) and also progressing by increments of 1. Table2 below provides further explication and examples. Table 2.... In PAGE 3: ... This basic coding structure reveals the similarities and differences between like systems. The coding structure for each language is supplemented with a legend, similar in format to Table2 , in which language-specific descriptions are listed. This combination leads to a thorough, accurate, and above all, consistent representation of formality inflection, one which lends itself to useful cross-linguistic comparisons.... ..."

Table 1: Number of vertices, examples, control transformations, and proxy vertices in our meshes. We show the computation time for our solver compared to that of Sumner and colleagues (MIK) [2005]. All timings are in seconds. Comparison of percent error distortion E between a full vertex representation and usage of proxy vertices shows that the latter is comparably accurate.

in Interactive Posing
by unknown authors
"... In PAGE 5: ... Resolution-Independent Interaction. Table1 provides statis- tics about our experiments, including the number of mesh vertices, examples, control transformations, and proxy vertices. The running times provided are given for one iteration of our nonlinear solver; all timing was measured on a 3.... In PAGE 5: ... 2005] requires approximately one sec- ond to generate new meshes from the set of human poses, whereas our system allows posing and shape blending more than an order of magnitude faster. We validate our approximations of the example meshes by treat- ing them as a mesh sequence and measuring error in terms of per- cent distortion [Karni and Gotsman 2004], shown in the last column of Table1 . Example reconstruction using deformation gradients for all vertices incurs error due to approximation with controls.... ..."

Table 1: Number of vertices, examples, control transformations, and proxy vertices in our meshes. We show the computation time for our solver compared to that of Sumner and colleagues (MIK) [2005]. All timings are in seconds. Comparison of percent error distortion E between a full vertex representation and usage of proxy vertices shows that the latter is comparably accurate.

in Interactive Posing
by unknown authors
"... In PAGE 5: ... Resolution-Independent Interaction. Table1 provides statis- tics about our experiments, including the number of mesh vertices, examples, control transformations, and proxy vertices. The running times provided are given for one iteration of our nonlinear solver; all timing was measured on a 3.... In PAGE 5: ... 2005] requires approximately one sec- ond to generate new meshes from the set of human poses, whereas our system allows posing and shape blending more than an order of magnitude faster. We validate our approximations of the example meshes by treat- ing them as a mesh sequence and measuring error in terms of per- cent distortion [Karni and Gotsman 2004], shown in the last column of Table1 . Example reconstruction using deformation gradients for all vertices incurs error due to approximation with controls.... ..."

Table 1: Number of vertices, examples, control transformations, and proxy vertices in our meshes. We show the computation time for our solver compared to that of Sumner and colleagues (MIK) [2005]. All timings are in seconds. Comparison of percent error distortion E between a full vertex representation and usage of proxy vertices shows that the latter is comparably accurate.

in Inverse kinematics for . . .
by Kevin G. Der, et al.
"... In PAGE 5: ... Resolution-Independent Interaction. Table1 provides statis- tics about our experiments, including the number of mesh vertices, examples, control transformations, and proxy vertices. The running times provided are given for one iteration of our nonlinear solver; all timing was measured on a 3.... In PAGE 5: ... 2005] requires approximately one sec- ond to generate new meshes from the set of human poses, whereas our system allows posing and shape blending more than an order of magnitude faster. We validate our approximations of the example meshes by treat- ing them as a mesh sequence and measuring error in terms of per- cent distortion [Karni and Gotsman 2004], shown in the last column of Table1 . Example reconstruction using deformation gradients for all vertices incurs error due to approximation with controls.... ..."

Table 2 Amplitude Perturbation Features

in Voice Pathology Assessment based on a Dialogue System and Speech Analysis
by Richard B. Reilly Φ, Rosalyn Moran Φ, Peter Lacy
"... In PAGE 2: ... A 20msecond epoch is necessary to give an accurate representation of pitch. Table 1 and Table2 provides a list of the twelve pitch and amplitude features employed in this study. No.... ..."

Table 6: Optimal performances for each model and representation using the TIMIT database.

in Parametric Subspace Modeling Of Speech Transitions
by K. Reinhard, M. Niranjan 1999
"... In PAGE 27: ...ecessary. A frame-wise comparison using our distance score is performed afterwards. Classi cation is done by nding the minimum score within each model. In Table6 we give the best results for each model and representation whereas a more accurate performance gures can be found in the tables mentioned above. We furthermore considered also for TIMIT a smoothing scheme which we... ..."
Cited by 8

Table 1: Summary of the eight input representations Plannett uses. Gauss uses only representation 4. Rep. Name Feats Features Included

in Human Expert-Level Performance on a Scientific Image Analysis Task by a System Using Combined Artificial Neural Networks
by Kevin Cherkauer 1996
"... In PAGE 2: ... Each of these subsets, or repre- sentations, was hand selected as a potentially sensible grouping of related features. The representations are summarized in Table1 . Representations 1{4 in the ta- ble concentrate on the rst six PCCs at di erent reso- lutions and the petal lter because these features work well with the Gaussian classi er.... In PAGE 2: ... Representations 1{4 in the ta- ble concentrate on the rst six PCCs at di erent reso- lutions and the petal lter because these features work well with the Gaussian classi er. The larger represen- tations 5{8 in Table1 were developed speci cally for the ANNs, because the ANNs tended to become more accurate as more features were added. For each of the eight representations, Plannett trains four ANNs that contain, respectively, 0, 5, 10, and 20 hidden units in one layer.... ..."
Cited by 57

Table 1: Summary of the eight input representations Plannett uses. Gauss uses only representation 4. Rep. Name Feats Features Included

in Human Expert-Level Performance on a Scientific Image Analysis Task by a System Using Combined Artificial Neural Networks
by Kevin Cherkauer 1996
"... In PAGE 2: ... Each of these subsets, or repre- sentations, was hand selected as a potentially sensible grouping of related features. The representations are summarized in Table1 . Representations 1{4 in the ta- ble concentrate on the rst six PCCs at di erent reso- lutions and the petal lter because these features work well with the Gaussian classi er.... In PAGE 2: ... Representations 1{4 in the ta- ble concentrate on the rst six PCCs at di erent reso- lutions and the petal lter because these features work well with the Gaussian classi er. The larger represen- tations 5{8 in Table1 were developed speci cally for the ANNs, because the ANNs tended to become more accurate as more features were added. For each of the eight representations, Plannett trains four ANNs that contain, respectively, 0, 5, 10, and 20 hidden units in one layer.... ..."
Cited by 57
Next 10 →
Results 1 - 10 of 11,247
Powered by: Apache Solr
  • About CiteSeerX
  • Submit and Index Documents
  • Privacy Policy
  • Help
  • Data
  • Source
  • Contact Us

Developed at and hosted by The College of Information Sciences and Technology

© 2007-2019 The Pennsylvania State University