Modern information retrieval systems, including web search, ads placement, and recommender systems, typically rely on learning from user feedback. Click models, which study how users interact with a ranked list of items, provide a useful understanding of user feedback for learning ranking models. Constructing "right" dependencies is the key of any successful click model. However, probabilistic graphical models (PGMs) have to rely on manually assigned dependencies, and oversimplify user behaviors. Existing neural network based methods promote PGMs by enhancing the expressive ability and allowing flexible dependencies, but still suffer from exposure bias and inferior estimation. In this paper, we propose a novel framework, Adversarial Imitation Click Model (AICM), based on imitation learning. Firstly, we explicitly learn the reward function that recovers users' intrinsic utility and underlying intentions. Secondly, we model user interactions with a ranked list as a dynamic system instead of one-step click prediction, alleviating the exposure bias problem. Finally, we minimize the JS divergence through adversarial training and learn a stable distribution of click sequences, which makes AICM generalize well across different distributions of ranked lists. A theoretical analysis has indicated that AICM reduces the exposure bias from $O(T^2)$ to $O(T)$. Our studies on a public web search dataset show that AICM not only outperforms state-of-the-art models in traditional click metrics but also achieves superior performance in addressing the exposure bias and recovering the underlying patterns of click sequences.
Because of the superior feature representation ability of deep learning, various deep Click-Through Rate (CTR) models are deployed in the commercial systems by industrial companies. To achieve better performance, it is necessary to train the deep CTR models on huge volume of training data efficiently, which makes speeding up the training process an essential problem. Different from the models with dense training data, the training data for CTR models is usually high-dimensional and sparse. To transform the high-dimensional sparse input into low-dimensional dense real-value vectors, almost all deep CTR models adopt the embedding layer, which easily reaches hundreds of GB or even TB. Since a single GPU cannot afford to accommodate all the embedding parameters, when performing distributed training, it is not reasonable to conduct the data-parallelism only. Therefore, existing distributed training platforms for recommendation adopt model-parallelism. Specifically, they use CPU (Host) memory of servers to maintain and update the embedding parameters and utilize GPU worker to conduct forward and backward computations. Unfortunately, these platforms suffer from two bottlenecks: (1) the latency of pull \& push operations between Host and GPU; (2) parameters update and synchronization in the CPU servers. To address such bottlenecks, in this paper, we propose the ScaleFreeCTR: a MixCache-based distributed training system for CTR models. Specifically, in SFCTR, we also store huge embedding table in CPU but utilize GPU instead of CPU to conduct embedding synchronization efficiently. To reduce the latency of data transfer between both GPU-Host and GPU-GPU, the MixCache mechanism and Virtual Sparse Id operation are proposed. Comprehensive experiments and ablation studies are conducted to demonstrate the effectiveness and efficiency of SFCTR.
Personalized recommendation system has become pervasive in various video platform. Many effective methods have been proposed, but most of them didn't capture the user's multi-level interest trait and dependencies between their viewed micro-videos well. To solve these problems, we propose a Self-over-Co Attention module to enhance user's interest representation. In particular, we first use co-attention to model correlation patterns across different levels and then use self-attention to model correlation patterns within a specific level. Experimental results on filtered public datasets verify that our presented module is useful.
Learning sophisticated feature interactions is crucial for Click-Through Rate (CTR) prediction in recommender systems. Various deep CTR models follow an Embedding & Feature Interaction paradigm. The majority focus on designing network architectures in Feature Interaction module to better model feature interactions while the Embedding module, serving as a bottleneck between data and Feature Interaction module, has been overlooked. The common methods for numerical feature embedding are Normalization and Discretization. The former shares a single embedding for intra-field features and the latter transforms the features into categorical form through various discretization approaches. However, the first approach surfers from low capacity and the second one limits performance as well because the discretization rule cannot be optimized with the ultimate goal of CTR model. To fill the gap of representing numerical features, in this paper, we propose AutoDis, a framework that discretizes features in numerical fields automatically and is optimized with CTR models in an end-to-end manner. Specifically, we introduce a set of meta-embeddings for each numerical field to model the relationship among the intra-field features and propose an automatic differentiable discretization and aggregation approach to capture the correlations between the numerical features and meta-embeddings. Comprehensive experiments on two public and one industrial datasets are conducted to validate the effectiveness of AutoDis over the SOTA methods.
Recently, deep learning-based models have been widely studied for click-through rate (CTR) prediction and lead to improved prediction accuracy in many industrial applications. However, current research focuses primarily on building complex network architectures to better capture sophisticated feature interactions and dynamic user behaviors. The increased model complexity may slow down online inference and hinder its adoption in real-time applications. Instead, our work targets at a new model training strategy based on knowledge distillation (KD). KD is a teacher-student learning framework to transfer knowledge learned from a teacher model to a student model. The KD strategy not only allows us to simplify the student model as a vanilla DNN model but also achieves significant accuracy improvements over the state-of-the-art teacher models. The benefits thus motivate us to further explore the use of a powerful ensemble of teachers for more accurate student model training. We also propose some novel techniques to facilitate ensembled CTR prediction, including teacher gating and early stopping by distillation loss. We conduct comprehensive experiments against 12 existing models and across three industrial datasets. Both offline and online A/B testing results show the effectiveness of our KD-based training strategy.
In many applications, such as recommender systems, online advertising, and product search, click-through rate (CTR) prediction is a critical task, because its accuracy has a direct impact on both platform revenue and user experience. In recent years, with the prevalence of deep learning, CTR prediction has been widely studied in both academia and industry, resulting in an abundance of deep CTR models. Unfortunately, there is still a lack of a standardized benchmark and uniform evaluation protocols for CTR prediction. This leads to the non-reproducible and even inconsistent experimental results among these studies. In this paper, we present an open benchmark (namely FuxiCTR) for reproducible research and provide a rigorous comparison of different models for CTR prediction. Specifically, we ran over 4,600 experiments for a total of more than 12,000 GPU hours in a uniform framework to re-evaluate 24 existing models on two widely-used datasets, Criteo and Avazu. Surprisingly, our experiments show that many models have smaller differences than expected and sometimes are even inconsistent with what reported in the literature. We believe that our benchmark could not only allow researchers to gauge the effectiveness of new models conveniently, but also share some good practices to fairly compare with the state of the arts. We will release all the code and benchmark settings.
Deep learning models in recommender systems are usually trained in the batch mode, namely iteratively trained on a fixed-size window of training data. Such batch mode training of deep learning models suffers from low training efficiency, which may lead to performance degradation when the model is not produced on time. To tackle this issue, incremental learning is proposed and has received much attention recently. Incremental learning has great potential in recommender systems, as two consecutive window of training data overlap most of the volume. It aims to update the model incrementally with only the newly incoming samples from the timestamp when the model is updated last time, which is much more efficient than the batch mode training. However, most of the incremental learning methods focus on the research area of image recognition where new tasks or classes are learned over time. In this work, we introduce a practical incremental method to train deep CTR models, which consists of three decoupled modules (namely, data, feature and model module). Our method can achieve comparable performance to the conventional batch mode training with much better training efficiency. We conduct extensive experiments on a public benchmark and a private dataset to demonstrate the effectiveness of our proposed method.
Tagging has been recognized as a successful practice to boost relevance matching for information retrieval (IR), especially when items lack rich textual descriptions. A lot of research has been done for either multi-label text categorization or image annotation. However, there is a lack of published work that targets at item tagging specifically for IR. Directly applying a traditional multi-label classification model for item tagging is sub-optimal, due to the ignorance of unique characteristics in IR. In this work, we propose to formulate item tagging as a link prediction problem between item nodes and tag nodes. To enrich the representation of items, we leverage the query logs available in IR tasks, and construct a query-item-tag tripartite graph. This formulation results in a TagGNN model that utilizes heterogeneous graph neural networks with multiple types of nodes and edges. Different from previous research, we also optimize both full tag prediction and partial tag completion cases in a unified framework via a primary-dual loss mechanism. Experimental results on both open and industrial datasets show that our TagGNN approach outperforms the state-of-the-art multi-label classification approaches.
Video moment retrieval aims to localize the target moment in an video according to the given sentence. The weak-supervised setting only provides the video-level sentence annotations during training. Most existing weak-supervised methods apply a MIL-based framework to develop inter-sample confrontment, but ignore the intra-sample confrontment between moments with semantically similar contents. Thus, these methods fail to distinguish the target moment from plausible negative moments. In this paper, we propose a novel Regularized Two-Branch Proposal Network to simultaneously consider the inter-sample and intra-sample confrontments. Concretely, we first devise a language-aware filter to generate an enhanced video stream and a suppressed video stream. We then design the sharable two-branch proposal module to generate positive proposals from the enhanced stream and plausible negative proposals from the suppressed one for sufficient confrontment. Further, we apply the proposal regularization to stabilize the training process and improve model performance. The extensive experiments show the effectiveness of our method. Our code is released at here.