Challenges
Ten self-paced challenges on synthetic data and on the Lab's research articles. Read, answer, get an instant check, and learn what each source does and does not show.
Your points and badges are kept only in this browser. There is no sign-in and no leaderboard. Clearing your browser data clears them.
0points earned
0of 10 challenges finished
0questions solved
3 challenges
4 challenges
3 challenges
Typed here only. It appears on the printed note and is not saved or sent anywhere.
Clears the points and answers stored in this browser. It does not delete anything you sent through the optional form.
Pick a level, or filter by topic and platform. Every challenge links to its article and its dataset page.
Read the evidence and count carefully. No prior parsing work needed. Hint costs: 1, 2 and 3 points.
Tell a factory reset from a SIM region change in a recovery log, link the two recovery files by run code, and say what a missing entry does not prove.
Find your way around a bug report: its 22 categories, anchor sections, how far back event sections reach, and what a cut section does and does not tell you.
Count wipes and region changes in plain-text recovery histories, notice what a 1970 clock and a cut file do to a count, and check your counts against a set with reference answers.
Compare versions and sources, and say what a result does and does not show. Hint costs: 2, 3 and 5 points.
Spot the three kinds of layout change (encoding, place, shape), tell an empty read from a missing feature, and use the habits that stop a reader from hiding the difference.
Report a setting that is not found as present, absent, unreadable or cannot say, and know why a missing key is not the same as a setting that is off.
Count and score a classifier on a synthetic set, spot the leakage column and the noise columns, and see why accuracy misleads when one class is rare.
Read the permission and behaviour profile of Android apps, build and score a baseline, and recognise leakage, lineage clusters, an ordinal column and benign look-alikes.
Reason with several stores, clocks and small samples at once. Hint costs: 3, 5 and 8 points.
Compare what a full file system extraction, a bug report and a sysdiagnose can say about a restart, a permission use, a reset or an app opening, how far back each reaches, and whose clock stamps the date.
Decide when records in different stores describe one event, count the witnesses behind a finding, and tell absent from not observable when a store is lost.
Turn timing and size behaviour into features, score a rare-class detector honestly, and recognise a leakage column and a shortcut rule that flatters a model.
No challenge matches these filters. Clear a filter to see more.
Each page links to the article and the dataset. Use a text editor, Python or a SQLite browser. All free.
Each question is checked on its own, straight away. Wrong answers cost nothing.
There are three hints per question. Each shows its cost before you open it, and a correct answer earns at least 1 point.
When every question is solved, the page shows what you should have learned.
Want to tell us who you are, or get Lab news by email? The form is optional: About you.
