On the representation and reuse of machine learning (ML) models
4 likes923 views
On the representation and reuse of machine learning models 1. The document discusses representing machine learning models in a generic way so they can be stored, shared, and deployed across different platforms and applications. 2. It proposes using the Predictive Model Markup Language (PMML) as a standard way to represent models that allows for "train once, deploy anywhere". 3. PMML provides a balance between being a generic representation that can be understood by any system, while also supporting more specific representations tailored to particular use cases or systems.
![PMML production: Scikit-Learn
from sklearn2pmml import sklearn2pmml
from sklearn2pmml.decoration import CategoricalDomain, ContinuousDomain
audit_df = pandas.read_csv("Audit.csv")
audit_mapper = DataFrameMapper([
(["Age", "Income", "Hours"], ContinuousDomain()),
(["Employment", "Education", "Marital", "Occupation"], [CategoricalDomain(), LabelBinarizer()]),
(["Gender", "Deductions"], [CategoricalDomain(), LabelEncoder()]),
("Adjusted", None)])
audit = audit_mapper.fit_transform(audit_df)
audit_classifier = DecisionTreeClassifier(min_samples_split = 10)
audit_classifier.fit(audit[:, 0:48], audit[:, 48].astype(int))
sklearn2pmml(audit_classifier, audit_mapper, "audit_tree.pmml")
16](https://image.slidesharecdn.com/ontherepresentationandreuseofmlmodels-170206154845/85/On-the-representation-and-reuse-of-machine-learning-ML-models-16-320.jpg)
![Comparison of feature spaces
R Scikit-Learn Apache Spark ML
Feature
identification
Named Positional Pseudo-named
Feature data type Any Float, Double Double
Feature operational
type
Continuous,
Categorical, Ordinal
Continuous Continuous,
pseudo-categorical
Dataset abstraction List<Map<String,?>> float[][] or double[][] List<double[]>
Effect of
transformations on
dataset size
Low Medium (sparse) to high (dense)
21](https://image.slidesharecdn.com/ontherepresentationandreuseofmlmodels-170206154845/85/On-the-representation-and-reuse-of-machine-learning-ML-models-21-320.jpg)
![Code generation
<TreeModel
missingValueStrategy="defaultChild"
>
<Node id="1" defaultChild="3">
<True/>
<Node id="2" score="0">
SimplePredicate: Age <= 52
ScoreDistribution:
"0" = 7, "1" = 0
</Node>
<Node id="3" score="1">
SimplePredicate: Age > 52
ScoreDistribution:
"0" = 1, "1" = 2
</Node>
</Node>
</TreeModel>
Object[] node_1(FieldValue age, ...){
if(age != null && age.asFloat() <= 52f){
return node_2(...);
} else {
return node_3(...);
}
}
Object[] node_2(...){
return new Object[]{"0", 7d, 0d};
}
Object[] node_3(...){
return new Object[]{"1", 1d, 2d};
}
Bad Idea!
29](https://image.slidesharecdn.com/ontherepresentationandreuseofmlmodels-170206154845/85/On-the-representation-and-reuse-of-machine-learning-ML-models-29-320.jpg)
On the representation and reuse of machine learning (ML) models
- 1. On the representation and reuse of machine learning models Villu Ruusmann Openscoring OÜ
- 3. Def: "Model" Output = func(Input) 3
- 5. The problem "Train once, deploy anywhere" 5
- 6. A solution Matching model representation (MR) with the task at hand: 1. Storing a generic and stable MR 2. Generating a wide variety of more specific and volatile MRs upon request 6
- 7. The Predictive Model Markup Language (PMML) ● XML dialect for marking up models and associated data transformations ● Version 1.0 in 1999, version 4.3 in 2016 ● "Conventions over configuration" ● 17 top-level model types + ensembling http://dmg.org/ http://dmg.org/pmml/pmml-v4-3.html http://dmg.org/pmml/products.html 7
- 8. A continuum from black to white boxes Introducing transparency in the form of rich, easy to use, well-documented APIs: 1. Unmarshalling and marshalling 2. Static analyses. Ex: schema querying 3. Dynamic analyses. Ex: scoring 4. Tracing and explaining individual predictions 8
- 9. The Zen of Machine Learning "Making the model requires large data and many cpus. Using it does not" --darren https://www.mail-archive.com/user@spark.apache.org/msg40636.html 9
- 10. Model training workflow Real-world feature space ML-platform feature space ML-platform model 10
- 11. Model deployment workflow Real-world feature space ML-platform feature space ML-platform model Real-world feature space Real-world model vs. 11
- 12. Model resources R code Scikit-Learn code Apache Spark ML code Original PMML markup Java code Python code Optimized PMML markup Training DeploymentVersioned storage 12
- 13. Comparison of model persistence options R Scikit-Learn Apache Spark ML Model data structure stability Fair to excellent Fair Poor Native serialization data format RDS (binary) Pickle (binary) SER (binary) and JSON (text) Export to PMML Few external N/A Built-in trait PMMLWritable Import from PMML Few external N/A JPMML projects JPMML-R and r2pmml JPMML-SkLearn and sklearn2pmml JPMML-SparkML (-Package) 13
- 14. PMML production: R library("r2pmml") auto <- read.csv("Auto.csv") auto$origin <- as.factor(auto$origin) auto.formula <- formula(mpg ~ (.) ^ 2 + # simple features and their two way interactions I(displacement / cylinders) + I(log(weight))) # derived features auto.lm <- lm(auto.formula, data = auto) r2pmml(auto.lm, "auto_lm.pmml", dataset = auto) auto.glm <- glm(auto.formula, data = auto, family = "gaussian") r2pmml(auto.glm, "auto_glm.pmml", dataset = auto) 14
- 15. R quirks ● No pipeline concept. Some workflow standardization efforts by third parties. Ex: caret package ● Many (equally right-) ways of doing the same thing. Ex: "formula interface" vs. "matrix interface" ● High variance in the design and quality of packages. Ex: academia vs. industry ● Model objects may enclose the training data set 15
- 16. PMML production: Scikit-Learn from sklearn2pmml import sklearn2pmml from sklearn2pmml.decoration import CategoricalDomain, ContinuousDomain audit_df = pandas.read_csv("Audit.csv") audit_mapper = DataFrameMapper([ (["Age", "Income", "Hours"], ContinuousDomain()), (["Employment", "Education", "Marital", "Occupation"], [CategoricalDomain(), LabelBinarizer()]), (["Gender", "Deductions"], [CategoricalDomain(), LabelEncoder()]), ("Adjusted", None)]) audit = audit_mapper.fit_transform(audit_df) audit_classifier = DecisionTreeClassifier(min_samples_split = 10) audit_classifier.fit(audit[:, 0:48], audit[:, 48].astype(int)) sklearn2pmml(audit_classifier, audit_mapper, "audit_tree.pmml") 16
- 17. Scikit-Learn quirks ● Completely schema-less at algorithm level. Ex: no identification of columns, no tracking of column groups ● Very limited, simple data structures. Mix of Python and C ● No built-in persistence mechanism. Serialization in generic pickle data format. Upon de-serialization, hope that class definitions haven't changed in the meantime. 17
- 18. PMML production: Apache Spark ML // $ spark-shell --packages org.jpmml:jpmml-sparkml-package:1.0-SNAPSHOT .. import org.jpmml.sparkml.ConverterUtil val df = spark.read.option("header", "true").option("inferSchema", "true").csv("Wine.csv") val formula = new RFormula().setFormula("quality ~ .") val regressor = new DecisionTreeRegressor() val pipeline = new Pipeline().setStages(Array(formula, regressor)) val pipelineModel = pipeline.fit(df) val pmmlBytes = ConverterUtil.toPMMLByteArray(df.schema, pipelineModel) Files.write(Paths.get("wine_tree.pmml"), pmmlBytes) 18
- 19. Apache Spark ML quirks ● Split schema. Static def via Dataset#schema(), dynamic def via Dataset column metadata ● Models make predictions in transformed output space ● High internal complexity, overhead. Ex: temporary Dataset columns for feature transformation ● Built-in PMML export capabilities leak the JPMML-Model library to application classpath 19
- 20. PMML consumption: Apache Spark ML // $ spark-submit --packages org.jpmml:jpmml-spark:1.0-SNAPSHOT .. import org.jpmml.spark.EvaluatorUtil; import org.jpmml.spark.TransformerBuilder; Evaluator evaluator = EvaluatorUtil.createEvaluator(new File("audit_tree.pmml")); TransformerBuilder pmmlTransformerBuilder = new TransformerBuilder(evaluator) .withLabelCol("Adjusted") // String column .withProbabilityCol("Adjusted_probability", Arrays.asList("0", "1")) // Vector column .exploded(true); Transformer pmmlTransformer = pmmlTransformerBuilder.build(); Dataset<Row> input = ...; Dataset<Row> output = pmmlTransformer.transform(input); 20
- 21. Comparison of feature spaces R Scikit-Learn Apache Spark ML Feature identification Named Positional Pseudo-named Feature data type Any Float, Double Double Feature operational type Continuous, Categorical, Ordinal Continuous Continuous, pseudo-categorical Dataset abstraction List<Map<String,?>> float[][] or double[][] List<double[]> Effect of transformations on dataset size Low Medium (sparse) to high (dense) 21
- 22. Feature declaration <DataField name="Age" dataType="float" optype="continuous"> <Interval closure="closedClosed" leftMargin="17.0" rightMargin="83.0"/> </DataField> <DataField name="Gender" dataType="string" optype="categorical"> <Value value="Male"/> <Value value="Female"/> <Value value="N/A" property="missing"/> </DataField> http://dmg.org/pmml/v4-3/DataDictionary.html <MiningField name="Age" outliers="asExtremeValues" lowValue="18.0" highValue="75.0"/> <MiningField name="Gender"/> http://dmg.org/pmml/v4-3/MiningSchema.html 22
- 23. Feature statistics <UnivariateStats field="Age"> <NumericInfo mean="38.30279" standardDeviation="13.01375" median="3.70"/> <ContStats> <Interval closure="openClosed" leftMargin="17.0" rightMargin="23.6"/> <!-- Intervals 2 through 9 omitted for clarity --> <Interval closure="openClosed" leftMargin="76.4" rightMargin="83.0"/> <Array type="int">261 360 297 340 280 156 135 51 13 6</Array> </ContStats> </UnivariateStats> <UnivariateStats field="Gender"> <Counts totalFreq="1899" missingFreq="0" invalidFreq="0"/> <DiscrStats> <Array type="string">Male Female</Array> <Array type="int">1307 592</Array> </DiscrStats> </UnivariateStats> http://dmg.org/pmml/v4-3/Statistics.html23
- 24. Comparison of tree models R Scikit-Learn Apache Spark ML Algorithms No built-in, many external Few built-in Single built-in Split type(s) Binary or multi-way; simple and derived features Binary; simple features Continuous features Rel. op. (<, <=) Rel. op. (<=) Categorical features Set op. (%in%) Pseudo-rel. op. (==) Pseudo-set op. Reuse Hard Easy to medium 24
- 25. Tree model declaration <TreeModel functionName="classification" splitCharacteristic="binarySplit"> <Node id="1" recordCount="165"> <True/> <Node id="2" score="1" recordCount="35"> <SimplePredicate field="Education" operator="equal" value="Master"/> <ScoreDistribution value="1" recordCount="25"/> <ScoreDistribution value="0" recordCount="10"/> </Node> <Node id="3" score="0" recordCount="130"> <SimplePredicate field="Education" operator="notEqual" value="Master"/> <ScoreDistribution value="1" recordCount="20"/> <ScoreDistribution value="0" recordCount="110"/> </Node> </Node> </TreeModel> http://dmg.org/pmml/v4-3/TreeModel.html 25
- 26. Optimization (1/3) <Node> SimplePredicate: Gender != "Male" <Node> SimplePredicate: Age <= 34.5 </Node> <Node> SimplePredicate: Age > 34.5 </Node> </Node> <Node> SimplePredicate: Gender == "Male" </Node> <Node> SimplePredicate: Gender == "Male" </Node> <Node> SimplePredicate: Age <= 34.5 </Node> <Node> SimplePredicate: Age > 34.5 </Node> Replacing "deep" binary splits with "shallow" multi-splits: 26
- 27. Optimization (2/3) <TreeModel noTrueChildStrategy="returnNullPrediction" > <Node> <True/> <Node score="4.333333333333333"> SimplePredicate: pH <= 2.93 </Node> <Node score="5.483870967741935"> SimplePredicate: pH > 2.93 </Node> </Node> </TreeModel> <TreeModel noTrueChildStrategy="returnLastPrediction" > <Node score="5.483870967741935"> <True/> <Node score="4.333333333333333"> SimplePredicate: pH <= 2.93 </Node> </Node> </TreeModel> Cutting the number of terminal nodes in half: 27
- 28. Optimization (3/3) <Node> SimplePredicate: Age > 35.5 <Node score="0" recordCount="40"> SimplePredicate: Income <= 194386 ScoreDistribution: "0" = 40, "1" = 0 </Node> <Node score="0" recordCount="8"> SimplePredicate: Income > 194386 ScoreDistribution: "0" = 7, "1" = 1 </Node> </Node> <Node score="0" recordCount="48"> SimplePredicate: Age > 35.5 ScoreDistribution: "0" = 47, "1" = 1 </Node> Removing split levels that don't affect the prediction: 28
- 29. Code generation <TreeModel missingValueStrategy="defaultChild" > <Node id="1" defaultChild="3"> <True/> <Node id="2" score="0"> SimplePredicate: Age <= 52 ScoreDistribution: "0" = 7, "1" = 0 </Node> <Node id="3" score="1"> SimplePredicate: Age > 52 ScoreDistribution: "0" = 1, "1" = 2 </Node> </Node> </TreeModel> Object[] node_1(FieldValue age, ...){ if(age != null && age.asFloat() <= 52f){ return node_2(...); } else { return node_3(...); } } Object[] node_2(...){ return new Object[]{"0", 7d, 0d}; } Object[] node_3(...){ return new Object[]{"1", 1d, 2d}; } Bad Idea! 29
- 30. Common pitfalls Not treating splits on continuous features with the required precision (tolerance < 0.5 ULP): ● Truncating values. Ex: "1.5000(..)1" → "1.50" ● Changing data type. Ex: float ↔ double ● Changing arithmetic expressions. Ex: (x1 / x2 ) ↔ (1 / x2 ) * x1 30
- 31. Comparison of regression models R Scikit-Learn Apache Spark ML Algorithms Few built-in, many external Many built-in Few built-in Term type(s) Simple and derived features; interactions Simple features Reuse Hard Easy 31
- 32. Regression model declaration <RegressionModel functionName="regression" normalizationMethod="none"> <RegressionTable intercept="15.50143741004145"> <!-- Simple continuous feature --> <NumericPredictor name="cylinders" coefficient="2.1609496766686194"/> <!-- Simple categorical feature --> <CategoricalPredictor name="origin" value="2" coefficient="-35.87525051244351"/> <CategoricalPredictor name="origin" value="3" coefficient="-38.206750156693424"/> <!-- Interaction --> <PredictorTerm coefficient="-0.007734946028064237"> <FieldRef field="cylinders"/> <FieldRef field="displacement"/> </PredictorTerm> <!-- Derived feature; I(log(weight)) is backed by a DerivedField element --> <NumericPredictor name="I(log(weight))" coefficient="4.874500863508498"/> </RegressionTable> </RegressionModel> http://dmg.org/pmml/v4-3/Regression.html32
- 33. Generalized regression model declaration <GeneralRegressionModel functionName="regression" linkFunction="identity"> <ParameterList> <Parameter name="p0" label="(intercept)"/> <Parameter name="p11" label="cylinders:displacement"/> </ParameterList> <PPMatrix> <PPCell value="1" predictorName="cylinders" parameterName="p11"/> <PPCell value="1" predictorName="displacement" parameterName="p11"/> </PPMatrix> <ParamMatrix> <PCell parameterName="p0" beta="15.50143741004145"/> <PCell parameterName="p11" beta="-0.007734946028064237"/> </ParamMatrix> </GeneralRegressionModel> http://dmg.org/pmml/v4-3/GeneralRegression.html 33
- 34. Code generation double mpg(FieldValue cylinders, FieldValue displacement, FieldValue weight, FieldValue origin){ return 15.50143741004145d + // intercept 2.1609496766686194d * cylinders.asDouble() + // simple continuous feature origin(origin.asString()) + // simple categorical feature -0.007734946028064237d * (cylinders.asDouble() * displacement.asDouble()) + // interaction 4.874500863508498d * Math.ln(weight.asDouble()) // derived feature } double origin(String origin){ switch(origin){ case "1": return 0d; // baseline case "2": return -35.87525051244351d; case "3": return -38.206750156693424d; } throw new IllegalArgumentException(origin); } 34
- 35. Grand summary R Scikit-Learn Apache Spark ML ML-platform model information density Medium to high Low Low to medium ML-platform model interpretability Good Bad Bad PMML markup optimizability Low High Medium ML-platform to PMML learning and migration effort Medium Low to medium Low 35