Collaborative Metric Learning

By Cheng-Kang Hsieh, Longqi Yang, Yin Cui, Tsung-Yi Lin, Serge J. Belongie, D. Estrin. The Web Conference, 2017

550 citations recorded by the source corpus.

Metric learning algorithms produce distance metrics that capture the important relationships among data. In this work, we study the connection between metric learning and collaborative filtering. We propose Collaborative Metric Learning (CML) which learns a joint metric space to encode not only users' preferences but also the user-user and item-item similarity. The proposed algorithm outperforms state-of-the-art collaborative filtering algorithms on a wide range of recommendation tasks and uncovers the underlying spectrum of users' fine-grained preferences. CML also achieves significant speedup for Top-K recommendation tasks using off-the-shelf, approximate nearest-neighbor search, with negligible accuracy reduction.

Subfields: Information Systems, Transportation, Signal Processing

View Paper: https://www.semanticscholar.org/paper/61a9f78bd184c0f95ffcbc10997d6b88a50be2c3

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Information Systems

In Transportation

In Signal Processing

Citations in the corpus

Identifying Ambiguous Similarity Conditions via Semantic Matching (2022, background): “The learned embeddings encode the relationship between objects well and facilitate downstream tasks [4, 16, 27, 31].”
A Survey on Accuracy-Oriented Neural Recommendation: From Collaborative Filtering to Information-Rich Recommendation (2021, background): “Distinct from CML that uses simple metric learning that assumes each user’s embedding is equally close to every item embedding she likes, LRML introduces the relation vectors r to capture the relationships between user and item pairs [47] .”
A review on deep learning for recommender systems: challenges and remedies (2018, background): “(2017), Hsieh et al. (2017) evaluate their models on both movie and book domains.”
Latent Relational Metric Learning via Memory-based Attention for Collaborative Ranking (2017, methodology): “Recently, Hseih et al. [8] revealed the potential implications pertaining to the usage of inner products to model user-item relationships.”
Distilling Structured Knowledge into Embeddings for Explainable and Accurate Recommendation (2019, background): “(In fact, the latter example is the main motivation for collaborative metric learning[12] where relevance transitivity is attempted to be maintained.”
Interest-aware Message-Passing GCN for Recommendation (2021, methodology): “Earlier research efforts mainly focus on the shallow models, such as BPR [27], CML [16], matrix factorization (MF) [19].”
Efficient and Explicit Modelling of Image Hierarchies for Image Restoration (2023, background): “And similarity propagation is related to the triangle inequality in a metric space [19, 27, 73].”
CLEAR: Contrastive-Prototype Learning with Drift Estimation for Resource Constrained Stream Mining (2021, background): “Contrastive Learning, also named as embedding learning, aims to learn a suitable embedding space [16, 24, 33, 40] to minimize the intra-class variation and maximize the inter-class variation.”
Neural Graph Collaborative Filtering (2019, background): “roduct can force user and item embeddings of an observed interaction close to each other, its linearity makes it insufficient to reveal the complex and nonlinear relationships between users and items [14, 15]. Towards this end, recent efforts [11, 14, 15, 35] focus on exploiting deep learning techni”
Deep Metric Learning with Angular Loss (2017, background): “…feature representation, deep metric learning has achieved promising results on various tasks, such as visual product search [1, 20, 17], face recognition [6, 30, 24], feature matching [7], ﬁne-grained image classiﬁcation [33, 38], zero-shot learning [11, 35] and collaborative ﬁltering [13].”
Deep Learning Based Recommender System (2017, background): “Collaborative Metric Learning (CML).”
UltraGCN: Ultra Simplification of Graph Convolutional Networks for Recommendation (2021, methodology): “…total, we compare UltraGCN with various types of the state-of-the-art models, covering MF-based (MF-BPR [15], ENMF [3]), metric learing-based (CML [12]), network embedding methods (DeepWalk [20], LINE [25], and Node2Vec [7]), and GCN-based (NGCF [27], NIA-GCN [24], LR-GCCF [4], LightGCN [10], and…”
Hierarchical Gating Networks for Sequential Recommendation (2019, background): “ple negative instances from missing data to learn the pair-wise user preference between positive and negative samples [29]. Recently, several works [11, 22] are proposed to weigh the missing data. In [14, 36], metric learning is applied to compute the distance between users and items. With the abili”
Translation-based Recommendation (2017, background): “As argued in [Chen et al. , 2012; Feng et al. , 2015; Hsieh et al. , 2017], the underlying metricity assumption brings better generalization ability.”
Decomposing fit semantics for product size recommendation in metric spaces (2018, background): “Recently, collaborative filtering was combined with metric learning and its effectiveness was shown on various recommendation tasks [3].”
Hierarchical Temporal Convolutional Networks for Dynamic Recommender Systems (2019, background): “Several approaches treat user interactions as static information by ignoring the temporal dimension, and try to learn a static similarity matrix between users and items [7, 11, 14, 26].”
Ranking Distillation: Learning Compact Ranking Models With High Performance for Recommender System (2018, methodology): “To get the student predicted rank for this document, we apply Weston et al [36]’s sequential sampling, and do it in a parallel manner [16].”