Learn Before
Candidate Improvements for an Underperforming Image Classifier
Possible improvements for an underperforming image classifier fall into two broad groups. Data-side options include collecting more labeled examples and making the training set more varied. Model-side options include training for more iterations, changing network capacity, adjusting regularization, or redesigning the architecture. Use evidence from the model's errors to decide which option to test rather than applying every change indiscriminately.
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
Choose the Right Improvement Path
Machine Learning Tasks Often Suggest Where to Focus Effort
Linking Project Goals to Engineering Decisions
Small Priority Shifts Can Change Team Throughput
Broad Machine Learning Guidance First, Deep Learning Guidance Later
Machine Learning Yearning @ DeepLearning.AI
Uncertainty About the Best Machine Learning Strategy Drives Experimentation
Candidate Improvements for an Underperforming Image Classifier
Learn After
A practical improvement for a weak cat detector
True or False: One suggested fix for an underperforming image classifier is to train it for more gradient-descent iterations.
To improve a model that misses rare cases, one option is to collect a more _____ training set, such as examples from unusual conditions.
A practical improvement for an underperforming image classifier
Trying a smaller neural network is one possible improvement for an underperforming message classifier.
To improve a spam filter, one idea is to train the model _____, by taking more optimization steps.
Match each road-sign classifier improvement idea to its description.
Put the steps in a sensible order when a spam filter is not accurate enough and the team wants to improve it.
Which option best describes collecting a more varied training set for a cat classifier?
Regularization is not one of the suggested ways to improve a cat detector that is performing poorly.
A weaker wildlife classifier may improve if you change the neural network _____, which means how its layers and hidden units are arranged.
Match each improvement type to the appropriate idea for a package detector.
Order the steps for improving a road sign detector by expanding the training data.
Improving an Underperforming Fruit Classifier
Improving an Image Classifier for Airport Luggage Checks
Broadening a Bicycle Detector Training Set