Choose a deep learning problem (binary classification, multiclass classification or regression)
You can use any dataset (including MNIST, Reuters, IMDB and Boston Housing Price).
Build a deep learning model.
Investigate different hyper-parameter settings.
Log your findings in a Jupyter notebook.
Convert your Jupyter notebook into a report. The report should be structured as follows
Introduction. State the purposes of your experiments
Methods. This where the workflow has happened.
Results. Graphs and tables, with commentary.
Evaluation. Discuss your findings and form a conclusion
Use markdown headings, subheadings, subsubheadings etc. to structure your report.
Credit will be awarded for:
report structure and quality as a document
the extent of the research
a clear objective
a systematic investigation
understanding of deep learning.
Extra credit will be given for:
clean, modular code
a dataset other than one mentioned above
extremely thorough examination of a model on a single dataset
consideration of statistical uncertainty when comparing hyper-parameter settings
comparison of a hypothesis space on different datasets of a similar type
any work or DL understanding that reaches beyond the topics and concepts discussed so far.
Zip your Jupyter notebook and any additional documents, for example, images
You must submit a zip file.
37 freelanceria on tarjonnut keskimäärin %project_bid_stats_avg_sub_26% %project_currencyDetails_sign_sub_27% tähän työhön
Hi, I have worked on ML projects on smart home energy usage & housing price prediction using Jupyter notebook. I would like to work on your project. Let me know if you want to discuss further. Regards, Monir