- We would prefer you to use Keras in this project due to its simplicity. However, any other framework is welcome.
- In these following challenges, we will provide you with training data which you can use to build your model. On the deadlines, we will publish our private test data for you to get the final score.
- You have to write a report per week in which you should describe/explain how you processed data, how you improved your model (using cross-validation, k-fold, grid-search or something like that) stage by stage.
- Create Trello cards right after you start doing challenges, set due dates and ask at least one of the leaders to confirm your attendance.
For example:
REPORT WEEK 1 CHALLENGE 1
| Date | Description | Top-3 acc (Training/Validation) | Top-1 acc (Training/Validation) |
|---|---|---|---|
| 12/02/2018 | First try with a basic SVM | 80/70 | 77.6/72.5 |
| 13/02/2018 | Using Lenet5 architecture | 85/84 | 80.5/82.4 |
| 14/02/2018 | It's Valentine's Day so I changed the number of units of all hidden layers to the number of days we have been together. | 99.99/50 | 90.1/19.2 |
| ... |
- Note that we will use top-1 and top-3 accuracy to evaluate your model. Please implement these metrics by yourself to make sure that the performance in your report is aligned with your final score.
In this challenge, you're aksed to create a model or using whatever AI algorithm you want to classify over 10 different breeds of dog. Follow this link to download the dataset.
Make sure you index the classes in the following way, otherwise your final score will be incorrect
| Class Name | Index |
|---|---|
| bullmastiff | 0 |
| chowchow | 1 |
| pug | 2 |
| maltese | 3 |
| huskysibir | 4 |
| dachshund | 5 |
| dalmatian | 6 |
| corgi | 7 |
| chihuahua | 8 |
| yorkshire | 9 |
- CNN network
- Building model with keras
- Transfer learning with keras tutorial (google yourself :D)
- 1st report: one week after you start doing challenge I
- 2nd report: two weeks after you start doing challenge I
- Private data: published after you submit 2nd report
The goal of this challenge is to develop a model that can detect the user's intention based on their query or input sentence. This is a basic sentence classification problem with 7 classes (7 intents), with around 2000 sentences each class in the training dataset. All the training data is contained in intent-train-data, there are 7 files respected to 7 classes. Each file contains all training examples of each class and one example (sentence) on each line. This dataset is extracted from nlu-benchmark
The 7 intents are: (Make sure you index the classes in the following order, otherwise your final score will be incorrect)
- SearchCreativeWork (e.g. Find me the I, Robot television show),
- GetWeather (e.g. Is it windy in Boston, MA right now?),
- BookRestaurant (e.g. I want to book a highly rated restaurant for me and my boyfriend tomorrow night),
- PlayMusic (e.g. Play the last track from Beyoncé off Spotify),
- AddToPlaylist (e.g. Add Diamonds to my roadtrip playlist)
- RateBook (e.g. Give 6 stars to Of Mice and Men)
- SearchScreeningEvent (e.g. Check the showtimes for Wonder Woman in Paris)
- 1st report: one week after you start doing challenge II
- 2nd report: two weeks after you start doing challenge II
- Private data: published after you submit 2nd report