Random competition: a simple, but efficient method for parallelizing inference systems
Abstract: "We present a very simple parallel execution model suitable for inference systems with nondeterministic choices (OR-branching points). The selection of OR-branches is done at random, with backtracking in case of failure. For parallelizing an inference system we employ a set of indepen...
Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Format: | Buch |
| Sprache: | English |
| Veröffentlicht: |
München
1990
|
| Schriftenreihe: | Technische Universität <München>: TUM
9050 |
| Schlagworte: | |
| Zusammenfassung: | Abstract: "We present a very simple parallel execution model suitable for inference systems with nondeterministic choices (OR-branching points). The selection of OR-branches is done at random, with backtracking in case of failure. For parallelizing an inference system we employ a set of independent processors, all of them solving an identical task. They only differ in the initialization of their random number generator (used for branch selection). Within this model, called random competition, we calculate analytically the parallel performance for arbitrary numbers of processors. This can be done without any experiments on a parallel machine As an application of this systematic approach we compute speedup expressions for specific problem classes defined by their run-time distributions. The results vary from a speedup of 1 for linearly degenerate search trees up to clearly "superlinear" speedup for strongly imbalanced search trees. Moreover, we are able to give estimates for the potential degree of OR-parallelism inherent in the different problem classes. Such an estimate is very important for the design of parallel inference machines. Finally, due to their simplicity, competition architectures are easy (and therefore low-priced) to build. |
| Beschreibung: | 14 S. |
Internformat
MARC
| LEADER | 00000nam a2200000 cb4500 | ||
|---|---|---|---|
| 001 | BV009224836 | ||
| 003 | DE-604 | ||
| 007 | t| | ||
| 008 | 940313s1990 xx |||| 00||| eng d | ||
| 035 | |a (OCoLC)27863620 | ||
| 035 | |a (DE-599)BVBBV009224836 | ||
| 040 | |a DE-604 |b ger |e rakddb | ||
| 041 | 0 | |a eng | |
| 049 | |a DE-29T | ||
| 100 | 1 | |a Ertel, Wolfgang |e Verfasser |4 aut | |
| 245 | 1 | 0 | |a Random competition |b a simple, but efficient method for parallelizing inference systems |c Wolfgang Ertel |
| 264 | 1 | |a München |c 1990 | |
| 300 | |a 14 S. | ||
| 336 | |b txt |2 rdacontent | ||
| 337 | |b n |2 rdamedia | ||
| 338 | |b nc |2 rdacarrier | ||
| 490 | 1 | |a Technische Universität <München>: TUM |v 9050 | |
| 520 | 3 | |a Abstract: "We present a very simple parallel execution model suitable for inference systems with nondeterministic choices (OR-branching points). The selection of OR-branches is done at random, with backtracking in case of failure. For parallelizing an inference system we employ a set of independent processors, all of them solving an identical task. They only differ in the initialization of their random number generator (used for branch selection). Within this model, called random competition, we calculate analytically the parallel performance for arbitrary numbers of processors. This can be done without any experiments on a parallel machine | |
| 520 | 3 | |a As an application of this systematic approach we compute speedup expressions for specific problem classes defined by their run-time distributions. The results vary from a speedup of 1 for linearly degenerate search trees up to clearly "superlinear" speedup for strongly imbalanced search trees. Moreover, we are able to give estimates for the potential degree of OR-parallelism inherent in the different problem classes. Such an estimate is very important for the design of parallel inference machines. Finally, due to their simplicity, competition architectures are easy (and therefore low-priced) to build. | |
| 650 | 4 | |a Künstliche Intelligenz | |
| 650 | 4 | |a Artificial intelligence | |
| 650 | 4 | |a Parallel processing (Electronic computers) | |
| 830 | 0 | |a Technische Universität <München>: TUM |v 9050 |w (DE-604)BV006185376 |9 9050 | |
| 943 | 1 | |a oai:aleph.bib-bvb.de:BVB01-006133990 | |
Datensatz im Suchindex
| _version_ | 1820868351391432704 |
|---|---|
| adam_text | |
| any_adam_object | |
| author | Ertel, Wolfgang |
| author_facet | Ertel, Wolfgang |
| author_role | aut |
| author_sort | Ertel, Wolfgang |
| author_variant | w e we |
| building | Verbundindex |
| bvnumber | BV009224836 |
| classification_rvk | SS 4637 |
| ctrlnum | (OCoLC)27863620 (DE-599)BVBBV009224836 |
| discipline | Informatik |
| format | Book |
| fullrecord | <?xml version="1.0" encoding="UTF-8"?><collection xmlns="http://www.loc.gov/MARC21/slim"><record><leader>00000nam a2200000 cb4500</leader><controlfield tag="001">BV009224836</controlfield><controlfield tag="003">DE-604</controlfield><controlfield tag="007">t|</controlfield><controlfield tag="008">940313s1990 xx |||| 00||| eng d</controlfield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(OCoLC)27863620</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-599)BVBBV009224836</subfield></datafield><datafield tag="040" ind1=" " ind2=" "><subfield code="a">DE-604</subfield><subfield code="b">ger</subfield><subfield code="e">rakddb</subfield></datafield><datafield tag="041" ind1="0" ind2=" "><subfield code="a">eng</subfield></datafield><datafield tag="049" ind1=" " ind2=" "><subfield code="a">DE-29T</subfield></datafield><datafield tag="100" ind1="1" ind2=" "><subfield code="a">Ertel, Wolfgang</subfield><subfield code="e">Verfasser</subfield><subfield code="4">aut</subfield></datafield><datafield tag="245" ind1="1" ind2="0"><subfield code="a">Random competition</subfield><subfield code="b">a simple, but efficient method for parallelizing inference systems</subfield><subfield code="c">Wolfgang Ertel</subfield></datafield><datafield tag="264" ind1=" " ind2="1"><subfield code="a">München</subfield><subfield code="c">1990</subfield></datafield><datafield tag="300" ind1=" " ind2=" "><subfield code="a">14 S.</subfield></datafield><datafield tag="336" ind1=" " ind2=" "><subfield code="b">txt</subfield><subfield code="2">rdacontent</subfield></datafield><datafield tag="337" ind1=" " ind2=" "><subfield code="b">n</subfield><subfield code="2">rdamedia</subfield></datafield><datafield tag="338" ind1=" " ind2=" "><subfield code="b">nc</subfield><subfield code="2">rdacarrier</subfield></datafield><datafield tag="490" ind1="1" ind2=" "><subfield code="a">Technische Universität <München>: TUM</subfield><subfield code="v">9050</subfield></datafield><datafield tag="520" ind1="3" ind2=" "><subfield code="a">Abstract: "We present a very simple parallel execution model suitable for inference systems with nondeterministic choices (OR-branching points). The selection of OR-branches is done at random, with backtracking in case of failure. For parallelizing an inference system we employ a set of independent processors, all of them solving an identical task. They only differ in the initialization of their random number generator (used for branch selection). Within this model, called random competition, we calculate analytically the parallel performance for arbitrary numbers of processors. This can be done without any experiments on a parallel machine</subfield></datafield><datafield tag="520" ind1="3" ind2=" "><subfield code="a">As an application of this systematic approach we compute speedup expressions for specific problem classes defined by their run-time distributions. The results vary from a speedup of 1 for linearly degenerate search trees up to clearly "superlinear" speedup for strongly imbalanced search trees. Moreover, we are able to give estimates for the potential degree of OR-parallelism inherent in the different problem classes. Such an estimate is very important for the design of parallel inference machines. Finally, due to their simplicity, competition architectures are easy (and therefore low-priced) to build.</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Künstliche Intelligenz</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Artificial intelligence</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Parallel processing (Electronic computers)</subfield></datafield><datafield tag="830" ind1=" " ind2="0"><subfield code="a">Technische Universität <München>: TUM</subfield><subfield code="v">9050</subfield><subfield code="w">(DE-604)BV006185376</subfield><subfield code="9">9050</subfield></datafield><datafield tag="943" ind1="1" ind2=" "><subfield code="a">oai:aleph.bib-bvb.de:BVB01-006133990</subfield></datafield></record></collection> |
| id | DE-604.BV009224836 |
| illustrated | Not Illustrated |
| indexdate | 2025-01-10T13:23:00Z |
| institution | BVB |
| language | English |
| oai_aleph_id | oai:aleph.bib-bvb.de:BVB01-006133990 |
| oclc_num | 27863620 |
| open_access_boolean | |
| owner | DE-29T |
| owner_facet | DE-29T |
| physical | 14 S. |
| publishDate | 1990 |
| publishDateSearch | 1990 |
| publishDateSort | 1990 |
| record_format | marc |
| series | Technische Universität <München>: TUM |
| series2 | Technische Universität <München>: TUM |
| spelling | Ertel, Wolfgang Verfasser aut Random competition a simple, but efficient method for parallelizing inference systems Wolfgang Ertel München 1990 14 S. txt rdacontent n rdamedia nc rdacarrier Technische Universität <München>: TUM 9050 Abstract: "We present a very simple parallel execution model suitable for inference systems with nondeterministic choices (OR-branching points). The selection of OR-branches is done at random, with backtracking in case of failure. For parallelizing an inference system we employ a set of independent processors, all of them solving an identical task. They only differ in the initialization of their random number generator (used for branch selection). Within this model, called random competition, we calculate analytically the parallel performance for arbitrary numbers of processors. This can be done without any experiments on a parallel machine As an application of this systematic approach we compute speedup expressions for specific problem classes defined by their run-time distributions. The results vary from a speedup of 1 for linearly degenerate search trees up to clearly "superlinear" speedup for strongly imbalanced search trees. Moreover, we are able to give estimates for the potential degree of OR-parallelism inherent in the different problem classes. Such an estimate is very important for the design of parallel inference machines. Finally, due to their simplicity, competition architectures are easy (and therefore low-priced) to build. Künstliche Intelligenz Artificial intelligence Parallel processing (Electronic computers) Technische Universität <München>: TUM 9050 (DE-604)BV006185376 9050 |
| spellingShingle | Ertel, Wolfgang Random competition a simple, but efficient method for parallelizing inference systems Technische Universität <München>: TUM Künstliche Intelligenz Artificial intelligence Parallel processing (Electronic computers) |
| title | Random competition a simple, but efficient method for parallelizing inference systems |
| title_auth | Random competition a simple, but efficient method for parallelizing inference systems |
| title_exact_search | Random competition a simple, but efficient method for parallelizing inference systems |
| title_full | Random competition a simple, but efficient method for parallelizing inference systems Wolfgang Ertel |
| title_fullStr | Random competition a simple, but efficient method for parallelizing inference systems Wolfgang Ertel |
| title_full_unstemmed | Random competition a simple, but efficient method for parallelizing inference systems Wolfgang Ertel |
| title_short | Random competition |
| title_sort | random competition a simple but efficient method for parallelizing inference systems |
| title_sub | a simple, but efficient method for parallelizing inference systems |
| topic | Künstliche Intelligenz Artificial intelligence Parallel processing (Electronic computers) |
| topic_facet | Künstliche Intelligenz Artificial intelligence Parallel processing (Electronic computers) |
| volume_link | (DE-604)BV006185376 |
| work_keys_str_mv | AT ertelwolfgang randomcompetitionasimplebutefficientmethodforparallelizinginferencesystems |