Use a Small Amount of Target-Distribution Data in Training with Auxiliary Data
When a model is being built for one target domain, the training set can include some examples from that same domain even if most of the training data comes from a different source. For example, if a wildlife monitor is trained mostly on broad online images but also includes a few hundred images from the specific reserve it will serve, then strong results on the reserve images in the training set but weaker results on reserve images in the training-dev set suggest that collecting more reserve images would likely help.
0
1
Tags
Machine Learning
Deep Learning
Supervised Learning
Dive into Deep Learning @ D2L
Data Science
Machine Learning Strategy
Machine Learning Yearning @ DeepLearning.AI
Related
Do Not Randomly Mix Sources When Building Validation and Test Sets
Use a Small Amount of Target-Distribution Data in Training with Auxiliary Data
When to Reduce the Influence of Extra Training Data
Training-Development Set
Comparing Errors Across Two Data Sources
Training–Dev Distribution Gap
Domain adaptation solves only certain distribution-shift problems
Choosing Development and Test Data for a Specific Application
Factory Photos and Inspection Tablet Images as a Distribution Mismatch Example
Random Train/Test Splits Can Be Misleading When Deployment Data Differs
Which data should define the dev and test sets for the plant-disease app?
Development and test examples should be drawn to match the kind of data the deployed system is expected to see.
Development and test sets should match the kind of data you expect later.
Match each data group for the plant-disease app with its role.
Order the dataset choices when training data and evaluation data come from different sources.
Why training data and evaluation data may come from different sources
Find the evaluation-set mismatch in a voice-command detector.
Why did strong desktop-photo results not predict mobile-upload success?
Choosing Training and Evaluation Data from Two Image Sources
If training uses extra web-sourced photos, the dev and test sets must also contain web-sourced photos.
Learn After
Testing a Data-Mismatch Hypothesis on a Shared Subset
A wildlife classifier has 8,000 camera-trap photos from reserve staff and 160,000 unrelated web photos. How should the reserve photos be used?
Combining In-Domain and Auxiliary Examples During Training
Adding target-distribution examples to training means the training set now includes data from the _____ distribution.
Match the training-set components with their distribution role in a model built for a specific target domain.
Order the steps for allocating product-review data between target and auxiliary sources.
Why can it help to include some target-distribution audio in both the training set and the training-dev set?
True or False: It can still be worthwhile to train on a dataset that comes from a different distribution than the dev/test set.
Strong results on source examples but weak results on deployment examples validate the hypothesis that more _____ data would help.
Match each training choice for a voice-command system to its likely effect.
Order the steps for training a defect classifier with mixed data sources.
Why Include a Small Amount of Target-Distribution Data in Training?
Using Target-Distribution Samples to Guide Training Decisions
Checking Whether More Same-Distribution Data Would Help