Collaborative Metric Learning

Hsieh, Cheng-Kang; Yang, Longqi; Cui, Yin; Lin, Tsung-Yi; Belongie, Serge; Estrin, Deborah

doi:10.1145/3038912.3052639

Cited by 461 publications

(440 citation statements)

References 36 publications

Supporting

Mentioning

436

Contrasting

Unclassified

Order By: Relevance

“…However, this is troublesome for the following two reasons. First, the sparse ground-truth labels {x u } treat all the unobserved (i.e., not clicked) items equally, while those unobserved items are by no means all irrelevant to that user [12] (otherwise, there is no point making recommendations). This makes it difficult, at least inefficient, for the embedding-based model to learn the underlying structure in data.…”

Section: Methodology 21 Problem Formulation and Preliminariesmentioning

confidence: 99%

Distilling Structured Knowledge into Embeddings for Explainable and Accurate Recommendation

Zhang

Xu²,

Zhou

et al. 2020

Proceedings of the 13th International Conference on Web Search and Data Mining

View full text Add to dashboard Cite

Recently, the embedding-based recommendation models (e.g., matrix factorization and deep models) have been prevalent in both academia and industry due to their effectiveness and flexibility. However, they also have such intrinsic limitations as lacking explainability and suffering from data sparsity. In this paper, we propose an end-to-end joint learning framework to get around these limitations without introducing any extra overhead by distilling structured knowledge from a differentiable path-based recommendation model. Through extensive experiments, we show that our proposed framework can achieve state-of-the-art recommendation performance and meanwhile provide interpretable recommendation reasons. CCS CONCEPTS• Information systems → Recommender systems; • Computing methodologies → Learning latent representations.

show abstract

Section: Methodology 21 Problem Formulation and Preliminariesmentioning

confidence: 99%

Distilling Structured Knowledge into Embeddings for Explainable and Accurate Recommendation

Zhang

Xu²,

Zhou

et al. 2020

Proceedings of the 13th International Conference on Web Search and Data Mining

View full text Add to dashboard Cite

show abstract

“…Conventional MFbased methods operate in inner product spaces, which are flexible but can easily overfit. To this end, CML [15] imposes the triangle inequality constraint, by adopting metric embeddings to represent users and items. Here the preference score is estimated by the negative ℓ 2 distance:…”

Section: Collaborative Metric Learning (Cml)mentioning

confidence: 99%

“…To achieve satisfactory performance in terms of both accuracy and efficiency, we propose to use the retrieved items as candidates, and adopt sophisticated ranking models to refine the results. As the preference ranking problem has been heavily studied [13,15,32], we employ existing models to study the effect of the CIGAR framework, and propose a candidate-oriented sampling strategy to further boost accuracy.…”

Section: Candidate-oriented Re-rankingmentioning

confidence: 99%

Candidate Generation with Binary Codes for Large-Scale Top-N Recommendation

Kang

McAuley

2019

Proceedings of the 28th ACM International Conference on Information and Knowledge Management

View full text Add to dashboard Cite

Generating the Top-N recommendations from a large corpus is computationally expensive to perform at scale. Candidate generation and re-ranking based approaches are often adopted in industrial settings to alleviate efficiency problems. However it remains to be fully studied how well such schemes approximate complete rankings (or how many candidates are required to achieve a good approximation), or to develop systematic approaches to generate high-quality candidates efficiently. In this paper, we seek to investigate these questions via proposing a candidate generation and re-ranking based framework (CIGAR), which first learns a preference-preserving binary embedding for building a hash table to retrieve candidates, and then learns to re-rank the candidates using real-valued ranking models with a candidate-oriented objective. We perform a comprehensive study on several large-scale real-world datasets consisting of millions of users/items and hundreds of millions of interactions. Our results show that CIGAR significantly boosts the Top-N accuracy against state-of-the-art recommendation models, while reducing the query time by orders of magnitude. We hope that this work could draw more attention to the candidate generation problem in recommender systems.

show abstract

“…where u i , v j are, respectively, user and item latent vectors in R d , B ⊂ S is the set of positive pairs in the considered mini-batch, N i j ⊂ {k |(i, k) S } is a set of negative samples per triplet, α > 0 is a margin constant, D is the Euclidean distance and w i j is a weight based on the number of negatives in N i j falling inside the α-ball to penalize items at a lower rank [20], [.] + = max(., 0), L c is regularization term (weighted by the hyper parameter λ c ) used to de-correlate the dimensions in the learned metric [7]. The recommendation for an user is then made by finding the k nearest items around her/him in the latent space.…”

Section: Collaborative Metric Learningmentioning

confidence: 99%

Improving Collaborative Metric Learning with Efficient Negative Sampling

Tran¹,

Hennequin²,

Royo-Letelier³

et al. 2019

Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval

View full text Add to dashboard Cite

Distance metric learning based on triplet loss has been applied with success in a wide range of applications such as face recognition, image retrieval, speaker change detection and recently recommendation with the Collaborative Metric Learning (CML) model. However, as we show in this article, CML requires large batches to work reasonably well because of a too simplistic uniform negative sampling strategy for selecting triplets. Due to memory limitations, this makes it difficult to scale in high-dimensional scenarios. To alleviate this problem, we propose here a 2-stage negative sampling strategy which finds triplets that are highly informative for learning. Our strategy allows CML to work effectively in terms of accuracy and popularity bias, even when the batch size is an order of magnitude smaller than what would be needed with the default uniform sampling. We demonstrate the suitability of the proposed strategy for recommendation and exhibit consistent positive results across various datasets.

show abstract

Collaborative Metric Learning

Cited by 461 publications

References 36 publications

Distilling Structured Knowledge into Embeddings for Explainable and Accurate Recommendation

Distilling Structured Knowledge into Embeddings for Explainable and Accurate Recommendation

Candidate Generation with Binary Codes for Large-Scale Top-N Recommendation

Improving Collaborative Metric Learning with Efficient Negative Sampling

Contact Info

Product

Resources

About