# Dataset Information

![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4133893%2F534e507c6bf938cf73300e54ccd6d807%2FScreenshot%202024-11-07%20at%2010.11.12AM.png?generation=1730954501154839&alt=media)

**ConCon** represents a learning task of two classes: **positive and negative**. All data instances are images based on the **CLEVR** framework. A ground truth rule can be used to determine the class affiliation of any image (green star in the attached figure). In order to correctly classify any image, learning this ground truth rule is sufficient. Since ConCon is a continual learning dataset, there are **three tasks** that are learned after one another. Each task is confounded by a **task-specific confounder** (spiral in the attached figure), i.e., a feature that is present in the positive but not in the negative images of that task. The task-specific confounder changes across tasks.

As can be seen in the attached figure, there are two dataset variants. There is the

- **Disjoint** dataset, where task-specific confounders **never appear in other tasks**, and the
- **Strict** dataset, where task-specific confounders **may appear in other tasks** as random features in both positive and negative samples.

We hypothesize that the main challenge of the disjoint dataset will be to identify that the actual ground truth features are not the ones that are easiest to learn. For the strict dataset, storing information from previous tasks such that the ground truth feature can be learned in new tasks appears to be important.

**We accept submissions on either dataset**, i.e., a submission is not required to perform well on both tasks. Submissions proposing solutions for both datasets are allowed but not required.