Existing research efforts for multi-interest candidate matching in recommender systems mainly focus on improving model architecture or incorporating additional information, neglecting the importance of training schemes. This work revisits the training framework and uncovers two major problems hindering the expressiveness of learned multi-interest representations. First, the current training objective (i.e., uniformly sampled softmax) fails to effectively train discriminative representations in a multi-interest learning scenario due to the severe increase in easy negative samples. Second, a routing collapse problem is observed where each learned interest may collapse to express information only from a single item, resulting in information loss. To address these issues, we propose the REMI framework, consisting of an Interest-aware Hard Negative mining strategy (IHN) and a Routing Regularization (RR) method. IHN emphasizes interest-aware hard negatives by proposing an ideal sampling distribution and developing a Monte-Carlo strategy for efficient approximation. RR prevents routing collapse by introducing a novel regularization term on the item-to-interest routing matrices. These two components enhance the learned multi-interest representations from both the optimization objective and the composition information. REMI is a general framework that can be readily applied to various existing multi-interest candidate matching methods. Experiments on three real-world datasets show our method can significantly improve state-of-the-art methods with easy implementation and negligible computational overhead. The source code will be released.
The Federated Learning (FL) paradigm is known to face challenges under heterogeneous client data. Local training on non-iid distributed data results in deflected local optimum, which causes the client models drift further away from each other and degrades the aggregated global model's performance. A natural solution is to gather all client data onto the server, such that the server has a global view of the entire data distribution. Unfortunately, this reduces to regular training, which compromises clients' privacy and conflicts with the purpose of FL. In this paper, we put forth an idea to collect and leverage global knowledge on the server without hindering data privacy. We unearth such knowledge from the dynamics of the global model's trajectory. Specifically, we first reserve a short trajectory of global model snapshots on the server. Then, we synthesize a small pseudo dataset such that the model trained on it mimics the dynamics of the reserved global model trajectory. Afterward, the synthesized data is used to help aggregate the deflected clients into the global model. We name our method Dynafed, which enjoys the following advantages: 1) we do not rely on any external on-server dataset, which requires no additional cost for data collection; 2) the pseudo data can be synthesized in early communication rounds, which enables Dynafed to take effect early for boosting the convergence and stabilizing training; 3) the pseudo data only needs to be synthesized once and can be directly utilized on the server to help aggregation in subsequent rounds. Experiments across extensive benchmarks are conducted to showcase the effectiveness of Dynafed. We also provide insights and understanding of the underlying mechanism of our method.
Non-IID data distribution across clients and poisoning attacks are two main challenges in real-world federated learning (FL) systems. While both of them have attracted great research interest with specific strategies developed, no known solution manages to address them in a unified framework. To universally overcome both challenges, we propose SmartFL, a generic approach that optimizes the server-side aggregation process with a small amount of proxy data collected by the service provider itself via a subspace training technique. Specifically, the aggregation weight of each participating client at each round is optimized using the server-collected proxy data, which is essentially the optimization of the global model in the convex hull spanned by client models. Since at each round, the number of tunable parameters optimized on the server side equals the number of participating clients (thus independent of the model size), we are able to train a global model with massive parameters using only a small amount of proxy data (e.g., around one hundred samples). With optimized aggregation, SmartFL ensures robustness against both heterogeneous and malicious clients, which is desirable in real-world FL where either or both problems may occur. We provide theoretical analyses of the convergence and generalization capacity for SmartFL. Empirically, SmartFL achieves state-of-the-art performance on both FL with non-IID data distribution and FL with malicious clients. The source code will be released.
Contrastive learning (CL) benefits the training of sequential recommendation models with informative self-supervision signals. Existing solutions apply general sequential data augmentation strategies to generate positive pairs and encourage their representations to be invariant. However, due to the inherent properties of user behavior sequences, some augmentation strategies, such as item substitution, can lead to changes in user intent. Learning indiscriminately invariant representations for all augmentation strategies might be sub-optimal. Therefore, we propose Equivariant Contrastive Learning for Sequential Recommendation (ECL-SR), which endows SR models with great discriminative power, making the learned user behavior representations sensitive to invasive augmentations (e.g., item substitution) and insensitive to mild augmentations (e.g., feature-level dropout masking). In detail, we use the conditional discriminator to capture differences in behavior due to item substitution, which encourages the user behavior encoder to be equivariant to invasive augmentations. Comprehensive experiments on four benchmark datasets show that the proposed ECL-SR framework achieves competitive performance compared to state-of-the-art SR models. The source code will be released.
Non-IID data distribution across clients and poisoning attacks are two main challenges in real-world federated learning systems. While both of them have attracted great research interest with specific strategies developed, no known solution manages to address them in a unified framework. To jointly overcome both challenges, we propose SmartFL, a generic approach that optimizes the server-side aggregation process with a small clean server-collected proxy dataset (e.g., around one hundred samples, 0.2% of the dataset) via a subspace training technique. Specifically, the aggregation weight of each participating client at each round is optimized using the server-collected proxy data, which is essentially the optimization of the global model in the convex hull spanned by client models. Since at each round, the number of tunable parameters optimized on the server side equals the number of participating clients (thus independent of the model size), we are able to train a global model with massive parameters using only a small amount of proxy data. We provide theoretical analyses of the convergence and generalization capacity for SmartFL. Empirically, SmartFL achieves state-of-the-art performance on both federated learning with non-IID data distribution and federated learning with malicious clients. The source code will be released.
Traffic forecasting is challenging due to dynamic and complicated spatial-temporal dependencies. However, existing methods still suffer from two critical limitations. Firstly, many approaches typically utilize static pre-defined or adaptively learned spatial graphs to capture dynamic spatial-temporal dependencies in the traffic system, which limits the flexibility and only captures shared patterns for the whole time, thus leading to sub-optimal performance. In addition, most approaches individually and independently consider the absolute error between ground truth and predictions at each time step, which fails to maintain the global properties and statistics of time series as a whole and results in trend discrepancy between ground truth and predictions. To this end, in this paper, we propose a Dynamic Adaptive and Adversarial Graph Convolutional Network (DAAGCN), which combines Graph Convolution Networks (GCNs) with Generative Adversarial Networks (GANs) for traffic forecasting. Specifically, DAAGCN leverages a universal paradigm with a gate module to integrate time-varying embeddings with node embeddings to generate dynamic adaptive graphs for inferring spatial-temporal dependencies at each time step. Then, two discriminators are designed to maintain the consistency of the global properties and statistics of predicted time series with ground truth at the sequence and graph levels. Extensive experiments on four benchmark datasets manifest that DAAGCN outperforms the state-of-the-art by average 5.05%, 3.80%, and 5.27%, in terms of MAE, RMSE, and MAPE, meanwhile, speeds up convergence up to 9 times. Code is available at https://github.com/juyongjiang/DAAGCN.
Session-based recommendation (SBR) aims to predict the user next action based on short and dynamic sessions. Recently, there has been an increasing interest in utilizing various elaborately designed graph neural networks (GNNs) to capture the pair-wise relationships among items, seemingly suggesting the design of more complicated models is the panacea for improving the empirical performance. However, these models achieve relatively marginal improvements with exponential growth in model complexity. In this paper, we dissect the classical GNN-based SBR models and empirically find that some sophisticated GNN propagations are redundant, given the readout module plays a significant role in GNN-based models. Based on this observation, we intuitively propose to remove the GNN propagation part, while the readout module will take on more responsibility in the model reasoning process. To this end, we propose the Multi-Level Attention Mixture Network (Atten-Mixer), which leverages both concept-view and instance-view readouts to achieve multi-level reasoning over item transitions. As simply enumerating all possible high-level concepts is infeasible for large real-world recommender systems, we further incorporate SBR-related inductive biases, i.e., local invariance and inherent priority to prune the search space. Experiments on three benchmarks demonstrate the effectiveness and efficiency of our proposal.
Session-based recommendation (SBR) aims to predict the user next action based on the ongoing sessions. Recently, there has been an increasing interest in modeling the user preference evolution to capture the fine-grained user interests. While latent user preferences behind the sessions drift continuously over time, most existing approaches still model the temporal session data in discrete state spaces, which are incapable of capturing the fine-grained preference evolution and result in sub-optimal solutions. To this end, we propose Graph Nested GRU ordinary differential equation (ODE), namely GNG-ODE, a novel continuum model that extends the idea of neural ODEs to continuous-time temporal session graphs. The proposed model preserves the continuous nature of dynamic user preferences, encoding both temporal and structural patterns of item transitions into continuous-time dynamic embeddings. As the existing ODE solvers do not consider graph structure change and thus cannot be directly applied to the dynamic graph, we propose a time alignment technique, called t-Alignment, to align the updating time steps of the temporal session graphs within a batch. Empirical results on three benchmark datasets show that GNG-ODE significantly outperforms other baselines.
Sequential recommendation (SR) aims to model users' dynamic preferences from their historical interactions. Recently, Transformer and convolution neural networks (CNNs) have shown great success in learning representations for SR. Nevertheless, Transformer mainly focus on capturing content-based global interactions, while CNNs effectively exploit local features in practical recommendation scenarios. Thus, how to effectively aggregate CNNs and Transformer to model both \emph{local} and \emph{global} dependencies of historical item sequence still remains an open challenge and is rarely studied in SR. To this regard, we inject locality inductive bias into Transformer by combining its global attention mechanism with a local convolutional filter, and adaptively determine the mixing importance on a personalized basis through a module and layer-aware adaptive mixture units, named AdaMCT. Moreover, considering that softmax-based attention may encourage unimodal activation, we introduce the Squeeze-Excitation Attention (with sigmoid activation) into sequential recommendation to capture multiple relevant items (keys) simultaneously. Extensive experiments on three widely used benchmark datasets demonstrate that AdaMCT significantly outperforms the previous Transformer and CNNs-based models by an average of 18.46% and 60.85% respectively in terms of NDCG@5 and achieves state-of-the-art performance.