Training Distilled Machine Learning Models


Ontology type: sgo:Patent     


Patent Info

DATE

N/A

AUTHORS

Vinyals, Oriol , DEAN, JEFFREY ADGATE , HINTON, Geoffrey E.

ABSTRACT

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a distilled machine learning model. One of the methods includes training a cumbersome machine learning model, wherein the cumbersome machine learning model is configured to receive an input and generate a respective score for each of a plurality of classes; and training a distilled machine learning model on a plurality of training inputs, wherein the distilled machine learning model is also configured to receive inputs and generate scores for the plurality of classes, comprising: processing each training input using the cumbersome machine learning model to generate a cumbersome target soft output for the training input; and training the distilled machine learning model to, for each of the training inputs, generate a soft output that matches the cumbersome target soft output for the training input. More... »

Related SciGraph Publications

  • 2000-12. Using a Neural Network to Approximate an Ensemble of Classifiers in NEURAL PROCESSING LETTERS
  • 2014-08. Mixture of experts: a literature survey in ARTIFICIAL INTELLIGENCE REVIEW
  • JSON-LD is the canonical representation for SciGraph data.

    TIP: You can open this SciGraph record using an external JSON-LD service: JSON-LD Playground Google SDTT

    [
      {
        "@context": "https://springernature.github.io/scigraph/jsonld/sgcontext.json", 
        "about": [
          {
            "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/2746", 
            "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
            "type": "DefinedTerm"
          }
        ], 
        "author": [
          {
            "name": "Vinyals, Oriol", 
            "type": "Person"
          }, 
          {
            "name": "DEAN, JEFFREY ADGATE", 
            "type": "Person"
          }, 
          {
            "name": "HINTON, Geoffrey E.", 
            "type": "Person"
          }
        ], 
        "citation": [
          {
            "id": "sg:pub.10.1007/s10462-012-9338-y", 
            "sameAs": [
              "https://app.dimensions.ai/details/publication/pub.1010470786", 
              "https://doi.org/10.1007/s10462-012-9338-y"
            ], 
            "type": "CreativeWork"
          }, 
          {
            "id": "sg:pub.10.1007/s10462-012-9338-y", 
            "sameAs": [
              "https://app.dimensions.ai/details/publication/pub.1010470786", 
              "https://doi.org/10.1007/s10462-012-9338-y"
            ], 
            "type": "CreativeWork"
          }, 
          {
            "id": "https://doi.org/10.1145/1150402.1150464", 
            "sameAs": [
              "https://app.dimensions.ai/details/publication/pub.1040632585"
            ], 
            "type": "CreativeWork"
          }, 
          {
            "id": "sg:pub.10.1023/a:1026530200837", 
            "sameAs": [
              "https://app.dimensions.ai/details/publication/pub.1047863261", 
              "https://doi.org/10.1023/a:1026530200837"
            ], 
            "type": "CreativeWork"
          }
        ], 
        "description": "

    Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a distilled machine learning model. One of the methods includes training a cumbersome machine learning model, wherein the cumbersome machine learning model is configured to receive an input and generate a respective score for each of a plurality of classes; and training a distilled machine learning model on a plurality of training inputs, wherein the distilled machine learning model is also configured to receive inputs and generate scores for the plurality of classes, comprising: processing each training input using the cumbersome machine learning model to generate a cumbersome target soft output for the training input; and training the distilled machine learning model to, for each of the training inputs, generate a soft output that matches the cumbersome target soft output for the training input.\n

    ", "id": "sg:patent.EP-2953066-A3", "keywords": [ "Learning", "method", "apparatus", "software", "computer storage", "machine", "cumbersome", "wherein", "input", "score", "plurality", "class", "processing", "output" ], "name": "TRAINING DISTILLED MACHINE LEARNING MODELS", "recipient": [ { "id": "https://www.grid.ac/institutes/grid.420451.6", "type": "Organization" } ], "sameAs": [ "https://app.dimensions.ai/details/patent/EP-2953066-A3" ], "sdDataset": "patents", "sdDatePublished": "2019-03-07T15:35", "sdLicense": "https://scigraph.springernature.com/explorer/license/", "sdPublisher": { "name": "Springer Nature - SN SciGraph project", "type": "Organization" }, "sdSource": "s3://com.uberresearch.data.dev.patents-pipeline/full_run_10/sn-export/5eb3e5a348d7f117b22cc85fb0b02730/0000100128-0000348334/json_export_b7bdf8d8.jsonl", "type": "Patent" } ]
     

    Download the RDF metadata as:  json-ld nt turtle xml License info

    HOW TO GET THIS DATA PROGRAMMATICALLY:

    JSON-LD is a popular format for linked data which is fully compatible with JSON.

    curl -H 'Accept: application/ld+json' 'https://scigraph.springernature.com/patent.EP-2953066-A3'

    N-Triples is a line-based linked data format ideal for batch operations.

    curl -H 'Accept: application/n-triples' 'https://scigraph.springernature.com/patent.EP-2953066-A3'

    Turtle is a human-readable linked data format.

    curl -H 'Accept: text/turtle' 'https://scigraph.springernature.com/patent.EP-2953066-A3'

    RDF/XML is a standard XML format for linked data.

    curl -H 'Accept: application/rdf+xml' 'https://scigraph.springernature.com/patent.EP-2953066-A3'


     

    This table displays all metadata directly associated to this object as RDF triples.

    54 TRIPLES      14 PREDICATES      30 URIs      21 LITERALS      2 BLANK NODES

    Subject Predicate Object
    1 sg:patent.EP-2953066-A3 schema:about anzsrc-for:2746
    2 schema:author N8f6d0c626efb4f8fbd00f0a0acd76fae
    3 schema:citation sg:pub.10.1007/s10462-012-9338-y
    4 sg:pub.10.1023/a:1026530200837
    5 https://doi.org/10.1145/1150402.1150464
    6 schema:description <p id="pa01" num="0001">Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a distilled machine learning model. One of the methods includes training a cumbersome machine learning model, wherein the cumbersome machine learning model is configured to receive an input and generate a respective score for each of a plurality of classes; and training a distilled machine learning model on a plurality of training inputs, wherein the distilled machine learning model is also configured to receive inputs and generate scores for the plurality of classes, comprising: processing each training input using the cumbersome machine learning model to generate a cumbersome target soft output for the training input; and training the distilled machine learning model to, for each of the training inputs, generate a soft output that matches the cumbersome target soft output for the training input. <img id="iaf01" file="imgaf001.tif" wi="85" he="77" img-content="drawing" img-format="tif"/></p>
    7 schema:keywords Learning
    8 apparatus
    9 class
    10 computer storage
    11 cumbersome
    12 input
    13 machine
    14 method
    15 output
    16 plurality
    17 processing
    18 score
    19 software
    20 wherein
    21 schema:name TRAINING DISTILLED MACHINE LEARNING MODELS
    22 schema:recipient https://www.grid.ac/institutes/grid.420451.6
    23 schema:sameAs https://app.dimensions.ai/details/patent/EP-2953066-A3
    24 schema:sdDatePublished 2019-03-07T15:35
    25 schema:sdLicense https://scigraph.springernature.com/explorer/license/
    26 schema:sdPublisher N318e33fcc1c64d19908de062f4fc1da5
    27 sgo:license sg:explorer/license/
    28 sgo:sdDataset patents
    29 rdf:type sgo:Patent
    30 N0855f5c4d84242328ca0c60c51abba3e schema:name Vinyals, Oriol
    31 rdf:type schema:Person
    32 N318e33fcc1c64d19908de062f4fc1da5 schema:name Springer Nature - SN SciGraph project
    33 rdf:type schema:Organization
    34 N45f59ff9b380473781596de39ee02ee4 rdf:first N82ce00ee469748d4bd239f9e6fba7b85
    35 rdf:rest Nd0dab339888b432790e5aa46d22ef440
    36 N82ce00ee469748d4bd239f9e6fba7b85 schema:name DEAN, JEFFREY ADGATE
    37 rdf:type schema:Person
    38 N8f6d0c626efb4f8fbd00f0a0acd76fae rdf:first N0855f5c4d84242328ca0c60c51abba3e
    39 rdf:rest N45f59ff9b380473781596de39ee02ee4
    40 Nd0dab339888b432790e5aa46d22ef440 rdf:first Ndc0694e17a3046708c3c45f20fb744b9
    41 rdf:rest rdf:nil
    42 Ndc0694e17a3046708c3c45f20fb744b9 schema:name HINTON, Geoffrey E.
    43 rdf:type schema:Person
    44 anzsrc-for:2746 schema:inDefinedTermSet anzsrc-for:
    45 rdf:type schema:DefinedTerm
    46 sg:pub.10.1007/s10462-012-9338-y schema:sameAs https://app.dimensions.ai/details/publication/pub.1010470786
    47 https://doi.org/10.1007/s10462-012-9338-y
    48 rdf:type schema:CreativeWork
    49 sg:pub.10.1023/a:1026530200837 schema:sameAs https://app.dimensions.ai/details/publication/pub.1047863261
    50 https://doi.org/10.1023/a:1026530200837
    51 rdf:type schema:CreativeWork
    52 https://doi.org/10.1145/1150402.1150464 schema:sameAs https://app.dimensions.ai/details/publication/pub.1040632585
    53 rdf:type schema:CreativeWork
    54 https://www.grid.ac/institutes/grid.420451.6 schema:Organization
     




    Preview window. Press ESC to close (or click here)


    ...