Matches in SemOpenAlex for { <https://semopenalex.org/work/W3169233287> ?p ?o ?g. }
- W3169233287 abstract "Recent advances in self-supervised learning with instance-level contrastive objectives facilitate unsupervised clustering. However, a standalone datum is not perceiving the context of the holistic cluster, and may undergo sub-optimal assignment. In this paper, we extend the mainstream contrastive learning paradigm to a cluster-level scheme, where all the data subjected to the same cluster contribute to a unified representation that encodes the context of each data group. Contrastive learning with this representation then rewards the assignment of each datum. To implement this vision, we propose twin-contrast clustering (TCC). We define a set of categorical variables as clustering assignment confidence, which links the instance-level learning track with the cluster-level one. On one hand, with the corresponding assignment variables being the weight, a weighted aggregation along the data points implements the set representation of a cluster. We further propose heuristic cluster augmentation equivalents to enable cluster-level contrastive learning. On the other hand, we derive the evidence lower-bound of the instance-level contrastive objective with the assignments. By reparametrizing the assignment variables, TCC is trained end-to-end, requiring no alternating steps. Extensive experiments show that TCC outperforms the state-of-the-art on challenging benchmarks." @default.
- W3169233287 created "2021-06-22" @default.
- W3169233287 creator A5009086593 @default.
- W3169233287 creator A5040995630 @default.
- W3169233287 creator A5042899882 @default.
- W3169233287 creator A5052917688 @default.
- W3169233287 creator A5061215231 @default.
- W3169233287 creator A5082634513 @default.
- W3169233287 date "2021-06-03" @default.
- W3169233287 modified "2023-10-16" @default.
- W3169233287 title "You Never Cluster Alone" @default.
- W3169233287 cites W1538131130 @default.
- W3169233287 cites W1663973292 @default.
- W3169233287 cites W1673310716 @default.
- W3169233287 cites W178113997 @default.
- W3169233287 cites W1821462560 @default.
- W3169233287 cites W1959608418 @default.
- W3169233287 cites W1987111416 @default.
- W3169233287 cites W2016381774 @default.
- W3169233287 cites W2020735245 @default.
- W3169233287 cites W2022686119 @default.
- W3169233287 cites W2095897464 @default.
- W3169233287 cites W2097645701 @default.
- W3169233287 cites W2099471712 @default.
- W3169233287 cites W2108501770 @default.
- W3169233287 cites W2110798204 @default.
- W3169233287 cites W2112796928 @default.
- W3169233287 cites W2117539524 @default.
- W3169233287 cites W2118858186 @default.
- W3169233287 cites W2127218421 @default.
- W3169233287 cites W2145094598 @default.
- W3169233287 cites W2165232124 @default.
- W3169233287 cites W2165874743 @default.
- W3169233287 cites W2187089797 @default.
- W3169233287 cites W2188365844 @default.
- W3169233287 cites W2194775991 @default.
- W3169233287 cites W2271840356 @default.
- W3169233287 cites W2293078015 @default.
- W3169233287 cites W2412589713 @default.
- W3169233287 cites W2533545350 @default.
- W3169233287 cites W2547875792 @default.
- W3169233287 cites W2548228487 @default.
- W3169233287 cites W2556467266 @default.
- W3169233287 cites W2593814746 @default.
- W3169233287 cites W2605021547 @default.
- W3169233287 cites W2730106296 @default.
- W3169233287 cites W2779692282 @default.
- W3169233287 cites W2780433069 @default.
- W3169233287 cites W2788552663 @default.
- W3169233287 cites W2798991696 @default.
- W3169233287 cites W2842511635 @default.
- W3169233287 cites W2883725317 @default.
- W3169233287 cites W2913668534 @default.
- W3169233287 cites W2913939497 @default.
- W3169233287 cites W2942985259 @default.
- W3169233287 cites W2949517790 @default.
- W3169233287 cites W2952433032 @default.
- W3169233287 cites W2953273646 @default.
- W3169233287 cites W2962852342 @default.
- W3169233287 cites W2962993015 @default.
- W3169233287 cites W2963557251 @default.
- W3169233287 cites W2963684088 @default.
- W3169233287 cites W2963761396 @default.
- W3169233287 cites W2964069537 @default.
- W3169233287 cites W2964074409 @default.
- W3169233287 cites W2964121744 @default.
- W3169233287 cites W2964160479 @default.
- W3169233287 cites W2968370607 @default.
- W3169233287 cites W2976847563 @default.
- W3169233287 cites W2986063762 @default.
- W3169233287 cites W2990500698 @default.
- W3169233287 cites W2993554215 @default.
- W3169233287 cites W2994560339 @default.
- W3169233287 cites W3005934118 @default.
- W3169233287 cites W3009561768 @default.
- W3169233287 cites W3012336650 @default.
- W3169233287 cites W3022061250 @default.
- W3169233287 cites W3034363127 @default.
- W3169233287 cites W3034561179 @default.
- W3169233287 cites W3034978746 @default.
- W3169233287 cites W3035524453 @default.
- W3169233287 cites W3047106524 @default.
- W3169233287 cites W3091642260 @default.
- W3169233287 cites W3095121901 @default.
- W3169233287 cites W3106709020 @default.
- W3169233287 cites W3110446398 @default.
- W3169233287 cites W3113415515 @default.
- W3169233287 cites W3115104721 @default.
- W3169233287 cites W3118608800 @default.
- W3169233287 cites W3121375627 @default.
- W3169233287 cites W3126337491 @default.
- W3169233287 cites W3135873104 @default.
- W3169233287 cites W3137513727 @default.
- W3169233287 cites W3154933889 @default.
- W3169233287 cites W3159901117 @default.
- W3169233287 cites W3163563550 @default.
- W3169233287 cites W3171153522 @default.
- W3169233287 cites W3199244260 @default.
- W3169233287 cites W2530816535 @default.
- W3169233287 cites W2551176409 @default.