0
$\begingroup$

I might be missing something obvious as I am new to machine learning. I am training an SSD Inception V2 for detecting buildings from satellite images. I use the Tensorflow Object Detection API. I am having troubles interpreting why the value of loss seems to change periodically:

loss

Please let me know if I need to add more information

Improve this question
$\endgroup$
2
  • 1
    $\begingroup$ Please provide more details about your model training. What hyperparameters you specified, batch size, optimization, etc. At the moment we don't have enough information to help you $\endgroup$
    – Leevo
    Nov 30 '20 at 13:35
  • $\begingroup$ Yes, please provide some more info about your data, model, and training hyperparameters. $\endgroup$
    – hH1sG0n3
    Nov 30 '20 at 15:50
1
$\begingroup$

Looks like you may be feeding training data in a very specific way or weights and biases reset after a specific period of time, e.g. end of epoch?

I would start with checking that

  • training and validation sets consist of a desired class ratio
  • datapoints within mini batches are being shuffled
  • model parameters are being correctly updated during training
Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.