diff --git a/project/MimaBuild.scala b/project/MimaBuild.scala
index adde213e361f0..fbf9b8e326d5c 100644
--- a/project/MimaBuild.scala
+++ b/project/MimaBuild.scala
@@ -88,7 +88,7 @@ object MimaBuild {
 
   def mimaSettings(sparkHome: File, projectRef: ProjectRef) = {
     val organization = "org.apache.spark"
-    val previousSparkVersion = "2.2.0"
+    val previousSparkVersion = "2.3.0"
     val project = projectRef.project
     val fullId = "spark-" + project + "_2.11"
     mimaDefaultSettings ++
diff --git a/project/MimaExcludes.scala b/project/MimaExcludes.scala
index b7e9cbc077c12..4246355f088eb 100644
--- a/project/MimaExcludes.scala
+++ b/project/MimaExcludes.scala
@@ -105,7 +105,50 @@ object MimaExcludes {
     ProblemFilters.exclude[InheritedNewAbstractMethodProblem]("org.apache.spark.ml.param.shared.HasValidationIndicatorCol.validationIndicatorCol"),
 
     // [SPARK-23042] Use OneHotEncoderModel to encode labels in MultilayerPerceptronClassifier
-    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.ml.classification.LabelConverter")
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.ml.classification.LabelConverter"),
+
+    // [SPARK-21842][MESOS] Support Kerberos ticket renewal and creation in Mesos
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.deploy.SparkHadoopUtil.getDateOfNextUpdate"),
+
+    // [SPARK-23366] Improve hot reading path in ReadAheadInputStream
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.io.ReadAheadInputStream.this"),
+
+    // [SPARK-22941][CORE] Do not exit JVM when submit fails with in-process launcher.
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.deploy.SparkSubmit.addJarToClasspath"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.deploy.SparkSubmit.mergeFileLists"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.deploy.SparkSubmit.prepareSubmitEnvironment$default$2"),
+
+    // Data Source V2 API changes
+    // TODO: they are unstable APIs and should not be tracked by mima.
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.ReadSupportWithSchema"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.SupportsScanColumnarBatch.createDataReaderFactories"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.SupportsScanColumnarBatch.createBatchDataReaderFactories"),
+    ProblemFilters.exclude[ReversedMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.SupportsScanColumnarBatch.planBatchInputPartitions"),
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.reader.SupportsScanUnsafeRow"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.DataSourceReader.createDataReaderFactories"),
+    ProblemFilters.exclude[ReversedMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.DataSourceReader.planInputPartitions"),
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.reader.SupportsPushDownCatalystFilters"),
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.reader.DataReader"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.SupportsReportStatistics.getStatistics"),
+    ProblemFilters.exclude[ReversedMissingMethodProblem]("org.apache.spark.sql.sources.v2.reader.SupportsReportStatistics.estimateStatistics"),
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.reader.DataReaderFactory"),
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.reader.streaming.ContinuousDataReader"),
+    ProblemFilters.exclude[MissingClassProblem]("org.apache.spark.sql.sources.v2.writer.SupportsWriteInternalRow"),
+    ProblemFilters.exclude[DirectMissingMethodProblem]("org.apache.spark.sql.sources.v2.writer.DataWriterFactory.createDataWriter"),
+    ProblemFilters.exclude[ReversedMissingMethodProblem]("org.apache.spark.sql.sources.v2.writer.DataWriterFactory.createDataWriter"),
+
+    // Changes to HasRawPredictionCol.
+    ProblemFilters.exclude[InheritedNewAbstractMethodProblem]("org.apache.spark.ml.param.shared.HasRawPredictionCol.rawPredictionCol"),
+    ProblemFilters.exclude[InheritedNewAbstractMethodProblem]("org.apache.spark.ml.param.shared.HasRawPredictionCol.org$apache$spark$ml$param$shared$HasRawPredictionCol$_setter_$rawPredictionCol_="),
+    ProblemFilters.exclude[InheritedNewAbstractMethodProblem]("org.apache.spark.ml.param.shared.HasRawPredictionCol.getRawPredictionCol"),
+
+    // [SPARK-15526][ML][FOLLOWUP] Make JPMML provided scope to avoid including unshaded JARs
+    (problem: Problem) => problem match {
+      case MissingClassProblem(cls) =>
+        !cls.fullName.startsWith("org.spark_project.jpmml") &&
+          !cls.fullName.startsWith("org.spark_project.dmg.pmml")
+      case _ => true
+    }
   )
 
   // Exclude rules for 2.3.x