, Tasks, Notes, Calendar), 100 task templates, and 520 task–application pairs across 52 Android applications.
Mobile GUI agents rarely transfer task success to new apps with the same goal
Authors introduce AnyAppBench to measure cross-application generalization and find consistent drops in performance and shifts in failure types when the app changes.
Top University
Tien Tran · Namho Koh · Daiki E. Matsunaga · Ayush Jain · Kee Eung Kim
KAIST · CMU
Research Digest··2 min read
Tran et al.
Why this paper
From KAIST and CMU
In one line
Mobile GUI agents do not generalize across apps for the same task; app-specific success does not transfer.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors
- ✓Reports numbers on named benchmarks
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§