Prototype

نموذج أوّلي

متجه قابل للتعلّم في فضاء التضمين يُمثّل مركز عنقود. في SwAV، تُربط السمات بمجموعة من النماذج الأوّلية المشتركة لتوليد تعيينات عنقدة ناعمة بدلاً من المقارنة الزوجية المباشرة.

A learnable vector in the embedding space representing a cluster center. In SwAV, features are matched against a shared set of prototypes to produce soft cluster assignments instead of direct pairwise comparison.

Also translated asنُموذَج بَدئي، متجه تمثيلي

First appears in this corpus in: SwAV: Unsupervised Learning of Visual Features by Contrasting Cluster Assignments (2020)

Appears in these papers

Emerging Properties in Self-Supervised Vision Transformers2021 in the sky ✦
SwAV: Unsupervised Learning of Visual Features by Contrasting Cluster Assignments2020 in the sky ✦
SwAV: Unsupervised Learning of Visual Features by Contrasting Cluster Assignments2020 in the sky ✦