Think Harder Where It Matters: Item-Level Adaptive Recursion for Sequential Recommendation

The paper introduces LoopRec, which recursively applies a single weight-shared block to sequential recommendation histories and uses a lightweight per-item gate to decide whether each item exits early or continues processing. The authors report that items in the same sequence need very different amounts of computation, and that non-uniform depth assignment outperforms fixed depth when given oracle access to the target item. Reported experiments cover four backbone families and public Amazon benchmarks, with LoopRec consistently beating matched fixed-depth counterparts. Learned allocations give deeper computation to lower-frequency and more recent items.

Why it matters

Suggests that uniform per-item compute in recommenders may waste capacity, and offers a way to learn depth allocation without auxiliary depth labels.

Read with care

Source claims not independently checked; the source text is a promotional post summarizing a paper and its reported benchmark gains.