Neural Compatibility Modeling with Attentive Knowledge Distillation

By Xuemeng Song, Fuli Feng, Xianjing Han, Xin Yang, Wei David Liu, Liqiang Nie. 2018

167 citations recorded by the source corpus.

Recently, the booming fashion sector and its huge potential benefits have attracted tremendous attention from many research communities. In particular, increasing research efforts have been dedicated to the complementary clothing matching as matching clothes to make a suitable outfit has become a daily headache for many people, especially those who do not have the sense of aesthetics. Thanks to the remarkable success of neural networks in various applications such as the image classification and speech recognition, the researchers are enabled to adopt the data-driven learning methods to analyze fashion items. Nevertheless, existing studies overlook the rich valuable knowledge (rules) accumulated in fashion domain, especially the rules regarding clothing matching. Towards this end, in this work, we shed light on the complementary clothing matching by integrating the advanced deep neural networks and the rich fashion domain knowledge. Considering that the rules can be fuzzy and different rules may have different confidence levels to different samples, we present a neural compatibility modeling scheme with attentive knowledge distillation based on the teacher-student network scheme. Extensive experiments on the real-world dataset show the superiority of our model over several state-of-the-art methods. Based upon the comparisons, we observe certain fashion insights that can add value to the fashion matching study. As a byproduct, we released the codes, and involved parameters to benefit other researchers.

Subfields: Computational Mechanics, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Computational Mechanics

1. David M. Beskow
2. Sungho Shin
3. Yoonho Boo
4. Liu Yang
5. Dacheng Tao
6. Bo Li
7. Tat-seng Chua
8. S. Maybank

In Computer Vision and Pattern Recognition

1. Xiangnan He
2. Tat-seng Chua
3. Xiang Wang
4. Dacheng Tao
5. Meng Wang
6. Liqiang Nie
7. Fuli Feng
8. Luc Van Gool

Citations in the corpus

• Knowledge Distillation: A Survey (2020, methodology): “…well reflect the neuron activations of convolutional neural network, some attention mechanisms are used in knowledge distillation to improve the performance of the student network (Zagoruyko and Komodakis 2017; Huang and Wang 2017; Srinivas and Fleuret 2018; Crowley et al. 2018; Song et al. 2018).”
• Empirical Analysis of Knowledge Distillation Technique for Optimization of Quantized Deep Neural Networks (2019, background): “Leveraging the knowledge contained in previously trained networks has attracted a lot of attention in a variety of applications for model compression (Tang et al., 2016; Song et al., 2018; Asami et al., 2017; Wang et al., 2019).”
• A Deep Look into Neural Ranking Models for Information Retrieval (2019, background): “These research works can be grouped into two categories: 1) learning with external structured knowledge such as knowledge bases [133, 29, 134, 135, 136, 137]; 2) learning with external unstructured knowledge such as retrieved top results, topics or tags [30, 138, 139].”
• Neural Compatibility Modeling With Probabilistic Knowledge Distillation (2020, methodology): “• AKD-DBPR : To better evaluate the knowledge encoding method, we introduced the baseline AKD-DBPR [2], which also utilizes the teacher-student scheme to encode the knowledge rules.”
• Explainable Outfit Recommendation with Joint Outfit Matching and Comment Generation (2018, methodology): “The top image encoder and the bottom image encoder are CNNs, which have been widely used in outfit recommendation [8], [13].”
• Finding and Characterizing Information Warfare Campaigns (2020, methodology): “Researchers have used DNNs for various vision tasks like the Imagenet Challenge [82, 143] and fashion recommendation [220].”
• Knowledge Distillation for Optimization of Quantized Deep Neural Networks (2019, background): “Leveraging the knowledge contained in previously trained networks has attracted attention in many applications for model compression [8, 9, 10, 11] and learning algorithms [12, 13, 14, 15].”
• Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networks (2021, background): “Feature maps and attention mechanisms have been demonstrated to be very useful in KD to supervise the training of student networks (Romero et al., 2015; Zagoruyko & Komodakis, 2017; Huang & Wang, 2017; Song et al., 2018; Ahn et al., 2019; Heo et al., 2019).”
• Neural Graph Collaborative Filtering (2019, background): “are many other forms of structural information can be useful for understanding user behaviors, such as the cross features [41] in context-aware and semantics-rich recommendation [22, 27], item knowledge graph [32], and social networks [34].”
• Adaptive multi-teacher multi-level knowledge distillation (2020, background): “Moreover, explicit knowledge from rules is incorporated into distillation learning through expectation regularization [9, 23].”