Filters








94,907 Hits in 7.7 sec

Contrastive Cross-domain Recommendation in Matching [article]

Ruobing Xie, Qi Liu, Liangdong Wang, Shukai Liu, Bo Zhang, Leyu Lin
2022 arXiv   pre-print
In this work, we propose a novel Contrastive Cross-Domain Recommendation (CCDR) framework for CDR in matching.  ...  contrastive learning (inter-CL) tasks for better representation learning and knowledge transfer.  ...  It is also convenient to extend the current positive samples 𝑒 𝑘 ∈ 𝑁 𝑡 𝑒 𝑖 to multi-hop neighbors for better generalization and diversity in CDR.  ... 
arXiv:2112.00999v2 fatcat:tmwwe3fo2zeknpevctteukgyx4

Bootstrap Latent Representations for Multi-modal Recommendation [article]

Xin Zhou, Hongyu Zhou, Yong Liu, Zhiwei Zeng, Chunyan Miao, Pengwei Wang, Yuan You, Feijun Jiang
2022 arXiv   pre-print
Moreover, existing multi-modal recommendation methods usually leverage randomly sampled negative examples in Bayesian Personalized Ranking (BPR) loss to guide the learning of user/item representations,  ...  BM3 alleviates both the need for contrasting with negative examples and the complex graph augmentation from an additional target network for contrastive view generation.  ...  The authors would like to express thanks to Professor Aixin Sun from Nanyang Technological University for his valuable suggestions.  ... 
arXiv:2207.05969v1 fatcat:kzsbdqc2oncvrlvuz22o64wecq

Multi-modal Graph Contrastive Learning for Micro-video Recommendation

Zixuan Yi, Xi Wang, Iadh Ounis, Craig Macdonald
2022 Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval  
To tackle this problem, we propose a novel learning method named Multi-Modal Graph Contrastive Learning (MMGCL), which aims to explicitly enhance multi-modal representation learning in a self-supervised  ...  Engagements in these platforms are facilitated by multi-modal recommendation systems.  ...  learning in top-𝑘 micro-video recommendation.  ... 
doi:10.1145/3477495.3532027 fatcat:44ybm5ouzndfldmicjvvt4tgyu

Re4: Learning to Re-contrast, Re-attend, Re-construct for Multi-interest Recommendation [article]

Shengyu Zhang, Lingxiao Yang, Dong Yao, Yujie Lu, Fuli Feng, Zhou Zhao, Tat-seng Chua, Fei Wu
2022 arXiv   pre-print
Since users' interests naturally exhibit multiple aspects, it is of increasing interest to develop multi-interest frameworks for recommendation, rather than represent each user with an overall embedding  ...  Specifically, Re4 encapsulates three backward flows, i.e., 1) Re-contrast, which drives each interest embedding to be distinct from other interests using contrastive learning; 2) Re-attend, which ensures  ...  Metrics are computed based on the top 20/50 recommended candidates (e.g., Recall@20).  ... 
arXiv:2208.08011v1 fatcat:sfh2hzdrknelpkaezmoib27eze

Re4: Learning to Re-contrast, Re-attend, Re-construct for Multi-interest Recommendation

Shengyu Zhang, Lingxiao Yang, Dong Yao, Yujie Lu, Fuli Feng, Zhou Zhao, Tat-seng Chua, Fei Wu
2022 Proceedings of the ACM Web Conference 2022  
Since users' interests naturally exhibit multiple aspects, it is of increasing interest to develop multi-interest frameworks for recommendation, rather than represent each user with an overall embedding  ...  Empirical studies validate that Re4 helps to learn learning distinct and effective multi-interest representations. CCS CONCEPTS • Information systems → Recommender systems.  ...  Metrics are computed based on the top 20/50 recommended candidates (e.g., Recall@20). Higher scores demonstrate better recommendation performance for all metrics.  ... 
doi:10.1145/3485447.3512094 fatcat:yfoyeulr3ffw7fppdzmljdbhnm

Improving Micro-video Recommendation via Contrastive Multiple Interests [article]

Beibei Li and Beihong Jin and Jiageng Song and Yisong Yu and Yiyuan Zheng and Wei Zhuo
2022 arXiv   pre-print
Moreover, it establishes the contrastive multi-interest loss to improve the robustness of interest embeddings and the performance of recommendations.  ...  Recently, contrastive learning provides a new opportunity for refining the existing recommendation techniques.  ...  2) We establish a multi-interest encoder based on implicit categories of items, and propose a contrastive multi-interest loss to minimize the difference between interests extracting from two augmented  ... 
arXiv:2205.09593v1 fatcat:ha5bntrhlrca7e4qyzoiwejs54

MIC: Model-agnostic Integrated Cross-channel Recommenders [article]

Yujie Lu, Ping Nie, Shengyu Zhang, Ming Zhao, Ruobing Xie, William Yang Wang, Yi Ren
2022 arXiv   pre-print
In this paper, we propose a model-agnostic integrated cross-channel (MIC) approach for the large-scale recommendation, which maximally leverages the inherent multi-channel mutual information to enhance  ...  Recent advances in this topic are based on multi-channel retrieval to efficiently measure users' interest on items from the massive candidate pool.  ...  Similarly, for user-user and item-item model, the loss function for a positive pair of examples ( ũ, 𝑢) and ( ṽ, 𝑣) is defined as: L 𝑢𝑢 = −log exp(sim(𝑢 𝑘 , ũ𝑘 )/𝜏) 𝑁 𝑗=1 𝑗≠𝑘 exp(sim(𝑢 𝑘  ... 
arXiv:2110.11570v2 fatcat:i52iepprnzdxtkc5sj4ja23jcy

Contrastive Learning for Debiased Candidate Generation in Large-Scale Recommender Systems [article]

Chang Zhou, Jianxin Ma, Jianwei Zhang, Jingren Zhou, Hongxia Yang
2021 arXiv   pre-print
We further improve upon CLRec and propose Multi-CLRec, for accurate multi-intention aware bias reduction.  ...  Based on the theoretical discovery, we design CLRec, a contrastive learning method to improve DCG in terms of fairness, effectiveness and efficiency in recommender systems with extremely large candidate  ...  MoCo (short for momentum contrast) [16] proposes a momentum method for updating the encoders based on the cached results and for stabilizing the training loss.  ... 
arXiv:2005.12964v9 fatcat:ju2bph7nqbfhhlvyswuexddova

Forest-based Deep Recommender

Chao Feng, Defu Lian, Zheng Liu, Xing Xie, Le Wu, Enhong Chen
2022 Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval  
To this end, we propose a Deep Forest-based Recommender (De-FoRec for short) for an efficient recommendation.  ...  The recently proposed tree-based deep recommendation models alleviate the problem by directly learning tree structure and representations under the guidance of recommendation objectives.  ...  Lastly, top-k items are recommended among the ranked items.  ... 
doi:10.1145/3477495.3531980 fatcat:2zgex3lmqrczpissfnf24lsqdq

Boost-RS: Boosted Embeddings for Recommender Systems and its Application to Enzyme-Substrate Interaction Prediction [article]

Xinmeng Li, Li-ping Liu, Soha Hassoun
2021 arXiv   pre-print
Recommender systems (RS), which are currently unexplored for the enzyme-substrate interaction prediction problem, can be utilized to provide enzyme recommendations for substrates, and vice versa.  ...  We show that each of our auxiliary tasks boosts learning of the embedding vectors, and that contrastive learning using Boost-RS outperforms attribute concatenation and multi-label learning.  ...  We then compare multi-label performance using contrastive loss (Boost-RS Multi-label), where we use weighted BCE for the multi-label loss, and using concatenation (NMF-Concat Multi-label), where we concatenate  ... 
arXiv:2109.14766v1 fatcat:gqeub2uhjzdh3p4nvqzrb2wfvi

Multi-granularity Item-based Contrastive Recommendation [article]

Ruobing Xie, Zhijie Qiu, Bo Zhang, Leyu Lin
2022 arXiv   pre-print
In this work, we propose a novel Multi-granularity item-based contrastive learning (MicRec) framework for the matching stage (i.e., candidate generation) in recommendation, which systematically introduces  ...  Contrastive learning (CL) has shown its power in recommendation.  ...  The top-k nearest items retrieved by the cosine similarities between title embeddings are viewed as the semantically similar positive samples in semantic-level item CL. • For taxonomies (e.g., tag), we  ... 
arXiv:2207.01387v1 fatcat:i6ky6csv6nhe5coa2qnxbjf7oe

Cross-domain User Preference Learning for Cold-start Recommendation [article]

Huiling Zhou, Jie Liu, Zhikang Li, Jin Yu, Hongxia Yang
2021 arXiv   pre-print
Cross-domain cold-start recommendation is an increasingly emerging issue for recommender systems.  ...  Now the method is serving online for the cross-domain cold micro-video recommendation.  ...  It can generalize user's diverse interest with top 𝐾 leaf-node representations based on his/her historical behaviors.  ... 
arXiv:2112.03667v1 fatcat:fvg2amg5qber7encbsxgosxgtu

On the Effectiveness of Sampled Softmax Loss for Item Recommendation [article]

Jiancan Wu, Xiang Wang, Xingyu Gao, Jiawei Chen, Hongcheng Fu, Tianyu Qiu, Xiangnan He
2022 arXiv   pre-print
Experimental results suggest that sampled softmax loss is more friendly to history and graph-based recommenders (e.g., SVD++ and LightGCN), but performs poorly for ID-based models (e.g., MF).  ...  In contrast, the history- and graph-based models, which naturally adjust representation magnitude according to node degree, are able to compensate for the shortcoming of sampled softmax loss.  ...  Multi-Sample Analysis of the Softmax Cross Entropy Loss for Learning-to-Rank with Binary based Contrastive Loss for Top-k Recommendation. CoRR abs/2109.00217 (2021). Relevance.  ... 
arXiv:2201.02327v1 fatcat:xxcxorondvbulge3q6pgfbgqee

Task-optimized User Clustering based on Mobile App Usage for Cold-start Recommendations

Bulou Liu, Bing Bai, Weibang Xie, Yiwen Guo, Hao Chen
2022 Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining  
Effective user clustering for article recommendations based on mobile app usage faces unique challenges, including (1) the gap between an active user's behavior of mobile app usage and article reading,  ...  To address the challenges, we propose a tailored Dual Alignment User Clustering (DAUC) model, which applies a sample-wise contrastive alignment to eliminate the gap between active users' mobile app usage  ...  contrastive loss to align the sample-wise representations.  ... 
doi:10.1145/3534678.3539105 fatcat:ectglcp54bhxnl7kznzq2hzczu

Multi-view Multi-behavior Contrastive Learning in Recommendation [article]

Yiqing Wu, Ruobing Xie, Yongchun Zhu, Xiang Ao, Xin Chen, Xu Zhang, Fuzhen Zhuang, Leyu Lin, Qing He
2022 arXiv   pre-print
In this work, we propose a novel Multi-behavior Multi-view Contrastive Learning Recommendation (MMCLR) framework, including three new CL tasks to solve the above challenges, respectively.  ...  Multi-behavior recommendation (MBR) aims to jointly consider multiple behaviors to improve the target behavior's performance.  ...  For HIT and NDCG, we report top 5 and 10; For each ground truth, we randomly sample 99 items that user did not interact with under the target behavior as negative samples.  ... 
arXiv:2203.10576v1 fatcat:xfdmok2qnjgbxolhgvz2xh3wl4
« Previous Showing results 1 — 15 out of 94,907 results