[madlib] branch master updated: clarify warm start vs transfer learning in user docs

[fm...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

fmcquillan pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/madlib.git


The following commit(s) were added to refs/heads/master by this push:
     new 87758d7  clarify warm start vs transfer learning in user docs
87758d7 is described below

commit 87758d701df0cba6f924fd1ff27f3ab931a4b8f8
Author: Frank McQuillan <fm...@pivotal.io>
AuthorDate: Fri Dec 20 16:53:26 2019 -0800

    clarify warm start vs transfer learning in user docs
---
 src/ports/postgres/modules/deep_learning/madlib_keras.sql_in   | 10 ++++++++--
 .../deep_learning/madlib_keras_fit_multiple_model.sql_in       | 10 ++++++++--
 2 files changed, 16 insertions(+), 4 deletions(-)

diff --git a/src/ports/postgres/modules/deep_learning/madlib_keras.sql_in b/src/ports/postgres/modules/deep_learning/madlib_keras.sql_in
index 0a395e8..1d2d0ba 100644
--- a/src/ports/postgres/modules/deep_learning/madlib_keras.sql_in
+++ b/src/ports/postgres/modules/deep_learning/madlib_keras.sql_in
@@ -1566,11 +1566,17 @@ metrics_iters             | {10}
 @anchor notes
 @par Notes
 
-1.  Refer to the deep learning section of the Apache MADlib
+1. Refer to the deep learning section of the Apache MADlib
 wiki [5] for important information including supported libraries
 and versions.
 
-2.  Classification is currently supported, not regression.
+2. Classification is currently supported, not regression.
+
+3. Reminder about the distinction between warm start and transfer learning.  Warm start uses model
+state (weights) from the model output table from a previous training run -
+set the 'warm_start' parameter to TRUE in the fit function.
+Transfer learning uses initial model state (weights) stored in the 'model_arch_table' - in this case set the
+'warm_start' parameter to FALSE in the fit function.
 
 @anchor background
 @par Technical Background
diff --git a/src/ports/postgres/modules/deep_learning/madlib_keras_fit_multiple_model.sql_in b/src/ports/postgres/modules/deep_learning/madlib_keras_fit_multiple_model.sql_in
index 669c5db..fbf3497 100644
--- a/src/ports/postgres/modules/deep_learning/madlib_keras_fit_multiple_model.sql_in
+++ b/src/ports/postgres/modules/deep_learning/madlib_keras_fit_multiple_model.sql_in
@@ -1308,11 +1308,17 @@ Note that the loss and accuracy values pick up from where the previous run left
 @anchor notes
 @par Notes
 
-1.  Refer to the deep learning section of the Apache MADlib
+1. Refer to the deep learning section of the Apache MADlib
 wiki [6] for important information including supported libraries
 and versions.
 
-2.  Classification is currently supported, not regression.
+2. Classification is currently supported, not regression.
+
+3. Reminder about the distinction between warm start and transfer learning.  Warm start uses model
+state (weights) from the model output table from a previous training run -
+set the 'warm_start' parameter to TRUE in the fit function.
+Transfer learning uses initial model state (weights) stored in the 'model_arch_table' - in this case set the
+'warm_start' parameter to FALSE in the fit function.
 
 @anchor background
 @par Technical Background