VBPR: Visual Bayesian Personalized Ranking from Implicit Feedback
By Ruining He, Julian McAuley. 2015
1,225 citations recorded by the source corpus.
Modern recommender systems model people and items by discovering or `teasing apart' the underlying dimensions that encode the properties of items and users' preferences toward them. Critically, such dimensions are uncovered based on user feedback, often in implicit form (such as purchase histories, browsing logs, etc.); in addition, some recommender systems make use of side information, such as product attributes, temporal information, or review text.However one important feature that is typically ignored by existing personalized recommendation and ranking methods is the visual appearance of the items being considered. In this paper we propose a scalable factorization model to incorporate visual signals into predictors of people's opinions, which we apply to a selection of large, real-world datasets. We make use of visual features extracted from product images using (pre-trained) deep networks, on top of which we learn an additional layer that uncovers the visual dimensions that best explain the variation in people's feedback. This not only leads to significantly more accurate personalized ranking methods, but also helps to alleviate cold start issues, and qualitatively to analyze the visual dimensions that influence people's opinions.
Subfields: Computer Vision and Pattern Recognition, Recommender Systems
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- Xiangnan He
- Dacheng Tao
- Tat-seng Chua
- Luc Van Gool
- Yi Yang
- Yongfeng Zhang
- Trevor Darrell
- Xiaogang Wang
In Recommender Systems
Citations in the corpus
- Ups and Downs: Modeling the Visual Evolution of Fashion Trends with One-Class Collaborative Filtering (2016, methodology): “Previous methods for ‘visually aware’ recommendation have made use of features from deep networks [12, 26] though made no use of temporal dynamics.”
- Ask the GRU: Multi-task Learning for Deep Text Recommendations (2016, methodology): “Finally, McAuley and Leskovec [4] used topic models and Almahairi et al. [24] used language models on review text.”
- Aspect-Aware Latent Factor Model: Rating Prediction with Ratings and Reviews (2018, background): “For example, various types of side information have been incorporated into MF to alleviate the cold-start problem, such as tags [30, 38], social relations [24, 34], reviews [23, 26, 39], and visual features [16].”
- Visually Explainable Recommendation (2018, methodology): “He et al [8] represented each product image as a fixed length vector, and infused it into the bayesian personalized ranking (BPR) framwork [24] to improve the performance of Top-N recommendation.”
- Neural Graph Collaborative Filtering (2019, methodology): “tasets in Table 1. Gowalla: This is the check-in dataset [21] obtained from Gowalla, where users share their locations by checking-in. To ensure the quality of the dataset, we use the 10-core setting [10], i.e., retaining users and items with at least ten interactions. Yelp2018∗: This dataset is ado”
- Joint Representation Learning for Top-N Recommendation with Heterogeneous Information Sources (2017, methodology): “Besides, they cover different information sources, including ratings (BPR), reviews (BPR-HFT, DeepCoNN), and images (VBPR, CKE), respectively.”
- Deep Item-based Collaborative Filtering for Top-N Recommendation (2018, background): “ the best single model in terms of fitting user ratings. SVD++ firstly factorizes the user-item rating matrix with implicit feedback [20] and is followed by lots of techniques for recommender systems [13,29,39]. Since user-item interactions on many recommender systems are based on implicit feedback ”
- A Survey on Session-based Recommender Systems (2019, background): “, users’ transaction behaviour data including view, click, add to cart and purchase, can be leveraged to learn users’ implicit general preference [35, 36].”
- Parallel Recurrent Neural Network Architectures for Feature-rich Session-based Recommendations (2016, methodology): “Image features that have been extracted using convolutional networks have been used in classical matrix factorization-type CF in [6, 16] to enhance the quality of recommendations.”
- Bootstrap Latent Representations for Multi-modal Recommendation (2022, methodology): “This setting has been widely used in previous studies [5], [14], [19], [42], [43].”
- Fusing Similarity Models with Markov Chains for Sparse Sequential Recommendation (2016, neutral): “[18, 28]), but also because it is a natural choice in our case as all comparison methods directly optimize this metric on the training set.”
- Personalized Fashion Recommendation with Visual Explanations based on Multimodal Attention Network: Towards Visually Explainable Recommendation (2019, methodology): “Following this character, recent years have witnessed many efforts on exploiting product images for fashion recommendation [12, 15, 33].”
- Artificial intelligence in recommender systems (2020, background): “Early works in other e-commerce recommendation areas take advantage of the features extracted from images using deep neural networks and integrate them with existing methods for clothing recommendation [194].”
- Neural Compatibility Modeling With Probabilistic Knowledge Distillation (2020, methodology): “Likewise, He and Mcauley [12] proposed a scalable matrix factorization approach that incorporates the visual features of product images to fulfil the recommendation task.”
- Graph-Refined Convolutional Network for Multimedia Recommendation with Implicit Feedback (2020, methodology): “Early works like VBPR [9] and ACF [4] integrate multimedia contents (e.”
- Reinforcement Knowledge Graph Reasoning for Explainable Recommendation (2019, methodology): “VBPR [8] is the Visual Bayesian Personalized Ranking method that builds upon the BPR model but incorporates visual product knowledge.”
- LLMRec: Large Language Models with Graph Augmentation for Recommendation (2023, methodology): “In contrast, approaches rely on limited content, such as VBPR [11] using only visual features, or CF-based architectures like NGCF [32], without side information, yield significantly diminished results.”