• 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 9,914
Next 10 →

Table 3. Best base algorithm results compared to best results after a 2nd iteration and radius=3 search. (HD=Hill Descent, SR=Successive Refinement, CN=Choose N Regions, RM=Run M Points).

in Automatic Creation of Domain-Specific Reconfigurable CPLDs for SoC
by Mark Holland, Scott Hauck 2005
"... In PAGE 5: ... If multiple algorithms found the same result, the algorithm that used the fewest runs is reported. As Table3 shows, running a second iteration of the algorithms was able to improve the area-delay product by up to .11x, with a mean area*delay gain of .... In PAGE 5: ...5x when compared to the base algorithms. Table3 shows that running a second iteration can be as effective as running a radial search, and it requires much less time. Also note that our base algorithms are performing reasonably well, as in all cases they are within .... In PAGE 5: ...18x of the best results we can easily find. Table3 shows that multiple algorithms provide high quality results. Considering the 2nd iteration column, the Run M Points algorithm actually matches the Choose N Regions algorithm for all but one of the data points (the other is only .... In PAGE 5: ...his rises to 5.6x to 11.9x. These are not necessarily the best fixed architectures for our set of domains, though. In fact, we have found the architectures that each domain prefers (in Table3 ), so it makes sense that these architectures might work well as fixed architectures. Table 4 shows the area-delay performance of each domain mapped to the architectures found using our best algorithm.... ..."
Cited by 3

Table 1. Example 1: Errors and rates

in FINITE ELEMENT APPROXIMATION OF THE NON-ISOTHERMAL STOKES-OLDROYD EQUATIONS
by Christopher Cox, Hyesuk Lee, David Szurley
"... In PAGE 13: ... We calculated errors on successively refined meshes. The convergence rate was then computed, and the results are shown in Table1 . The table shows that our compu-... ..."

Table 2: Refined instances of the RSP model

in Threading by Appointment
by Christoph M. Kirsch 2004
"... In PAGE 6: ...Table 3: Real-time instances of the RSP model addition, SEDA uses multiple stages of event-processing machinery that can be modeled by multiple RSPs. Table2 relates Capriccio and SEDA in terms of the RSP model, suggesting that recent implementa- tions of thread-based and event-driven models become more and more alike. There are also hybrids of thread-based and event-driven models that have successfully been implemented, e.... ..."

Table 5: MT output examples from the BTEC test set before and after automatic refinements applied to the grammar and lexicon.

in Improving a Transfer-Based MT System with Automatic Refinements
by Ariadna Font Llitjós, Jaime Carbonell, Alon Lavie 2007
"... In PAGE 7: ... 6 Language Independence of Approach In 56 cases, the additional generation capabili- ties of the refined system successfully produced a better translation than the baseline system; 37 of these improvements were ranked as 1-best by the decoder. Table5 shows examples of the three most common types of fixes yielded by automatic re- finements. The same mechanisms to automatically extend and refine an English to Spanish TBMT system are valid when applied to a very different language pair, i.... ..."
Cited by 1

Table 10 Comparison of integration maturity items

in INFORMATION SYSTEMS MANAGEMENT MATURITY AND
by Garry Spicer, Garry D. Spicer, Garry Spicer, Michael Basil Ph. D
"... In PAGE 6: ...able 9 Comparison of organization maturity items ....................................................... 36 Table10 Comparison of integration maturity items.... In PAGE 88: ... The study described here also successfully refined the items originally used by Karimi et al. (1996) (see Table 7 to Table10 ), and added an item to the organization component of IS maturity (the influence of IT personnel on IT planning and implementation, see Appendix A, Q28). This investigation successfully refined the IT security effectiveness measurement introduced by Kankanhalli et al.... ..."

Table 2: Effect comparison between the manual refinement interface and the term feedback interface

in ABSTRACT User Term Feedback in Interactive Text-based Image Retrieval
by unknown authors
"... In PAGE 3: ... The undesired terms are the query terms that do not appear in the target descriptions. Table2 shows the average percentage of desired or undesired query terms (with respect to the total number of terms in the description of the target image) that are specified among all iterations through each interface. Our results indicate that the percentage of desired terms is Table 1: Overall performance of two interfaces Manual refinement Term feedback Success Rate 0.... ..."

Table 3 Analysis of type 1 and type 2 errors

in 1 2 3 4 5 6 7 8
by unknown authors
"... In PAGE 6: ... System errors can be minimized by on-going efforts to successively refine the knowledge base. Overall, there are two major types of impact errors associated with the system ( Table3 ). Type 1 errors are false negatives and Type 2 errors are false positives.... ..."

Table 1 presents the progression of grid sizes through five levels of refinement. The computational mesh after the second refinement is shown in Fig. 2. For reference purposes, the original serial adaptation code consists of approx- imately 1,300 lines of C and requires 6.4 seconds to execute this simulation on a 250 MHz MIPS R10000 processor. Note that a flow solver was not run between between successive adaptations.

in Parallelization of a Dynamic Unstructured Application using Three Leading Paradigms
by Leonid Oliker, Rupak Biswas 1999
"... In PAGE 3: ... Table1 : Progression of grid sizes through five levels of adaptation.... In PAGE 5: ... Several options may be set within PLUM, including predictive or non-predictive refinement, global or diffusive parti- tioning, and synchronous or asynchronous communication. Tables 2 and 3 present the results for the best combination of these options on a T3E and an Origin2000, through the five refinement steps shown in Table1 . Note that results are not presented for less than eight processors of the T3E because of memory constraints.... ..."
Cited by 10

Table 1 presents the progression of grid sizes through five levels of refinement. The computational mesh after the second refinement is shown in Fig. 2. For reference purposes, the original serial adaptation code consists of approx- imately 1,300 lines of C and requires 6.4 seconds to execute this simulation on a 250 MHz MIPS R10000 processor. Note that a flow solver was not run between between successive adaptations.

in Parallelization of a Dynamic Unstructured Application using Three Leading Paradigms
by Leonid Oliker National, Leonid Oliker 1999
"... In PAGE 3: ... Table1 : Progression of grid sizes through five levels of adaptation.... In PAGE 5: ... Several options may be set within PLUM, including predictive or non-predictive refinement, global or diffusive parti- tioning, and synchronous or asynchronous communication. Tables 2 and 3 present the results for the best combination of these options on a T3E and an Origin2000, through the five refinement steps shown in Table1 . Note that results are not presented for less than eight processors of the T3E because of memory constraints.... ..."
Cited by 10

Table 1. Comparison of BoostMap, FastMap and using brute-force search, for the purpose of retrieving the exact nearest neighbors successfully for 95% or 100% of the queries, using filter-and-refine retrieval DX # per query is the total number of DX computations needed per query, in order to embed the query and rank the top p candidates selected at the filter step. The exact DX column shows the results for brute-force search, in which we simply evaluate DX distances between the query and all database images.

in Filtering Methods for Similarity-Based Multimedia Retrieval
by unknown authors
"... In PAGE 7: ... The reason that we want embeddings of higher quality is that those embeddings will allow us to achieve the desired accuracy while selecting a smaller number of objects at the filter step, thus reducing the processing time for the refine step. Table1 shows the results of BoostMap vs. FastMap.... ..."
Next 10 →
Results 1 - 10 of 9,914
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