To date, various neural methods have been proposed for causal effect estimation based on observational data, where a default assumption is the same distribution and availability of variables at both training and inference (i.e., runtime) stages. However, distribution shift (i.e., domain shift) could happen during runtime, and bigger challenges arise from the impaired accessibility of variables. This is commonly caused by increasing privacy and ethical concerns, which can make arbitrary variables unavailable in the entire runtime data and imputation impractical. We term the co-occurrence of domain shift and inaccessible variables runtime domain corruption, which seriously impairs the generalizability of a trained counterfactual predictor. To counter runtime domain corruption, we subsume counterfactual prediction under the notion of domain adaptation. Specifically, we upper-bound the error w.r.t. the target domain (i.e., runtime covariates) by the sum of source domain error and inter-domain distribution distance. In addition, we build an adversarially unified variational causal effect model, named VEGAN, with a novel two-stage adversarial domain adaptation scheme to reduce the latent distribution disparity between treated and control groups first, and between training and runtime variables afterwards. We demonstrate that VEGAN outperforms other state-of-the-art baselines on individual-level treatment effect estimation in the presence of runtime domain corruption on benchmark datasets.
To address privacy concerns and reduce network latency, there has been a recent trend of compressing cumbersome recommendation models trained on the cloud and deploying compact recommender models to resource-limited devices for real-time recommendation. Existing solutions generally overlook device heterogeneity and user heterogeneity. They either require all devices to share the same compressed model or the devices with the same resource budget to share the same model. However, even users with the same devices may have different preferences. In addition, they assume the available resources (e.g., memory) for the recommender on a device are constant, which is not reflective of reality. In light of device and user heterogeneities as well as dynamic resource constraints, this paper proposes a Personalized Elastic Embedding Learning framework (PEEL) for on-device recommendation, which generates personalized embeddings for devices with various memory budgets in once-for-all manner, efficiently adapting to new or dynamic budgets, and effectively addressing user preference diversity by assigning personalized embeddings for different groups of users. Specifically, it pretrains using user-item interaction instances to generate the global embedding table and cluster users into groups. Then, it refines the embedding tables with local interaction instances within each group. Personalized elastic embedding is generated from the group-wise embedding blocks and their weights that indicate the contribution of each embedding block to the local recommendation performance. PEEL efficiently generates personalized elastic embeddings by selecting embedding blocks with the largest weights, making it adaptable to dynamic memory budgets. Extensive experiments are conducted on two public datasets, and the results show that PEEL yields superior performance on devices with heterogeneous and dynamic memory budgets.
Multi-modal knowledge graph (MKG) includes triplets that consist of entities and relations and multi-modal auxiliary data. In recent years, multi-hop multi-modal knowledge graph reasoning (MMKGR) based on reinforcement learning (RL) has received extensive attention because it addresses the intrinsic incompleteness of MKG in an interpretable manner. However, its performance is limited by empirically designed rewards and sparse relations. In addition, this method has been designed for the transductive setting where test entities have been seen during training, and it works poorly in the inductive setting where test entities do not appear in the training set. To overcome these issues, we propose TMR (Topology-aware Multi-hop Reasoning), which can conduct MKG reasoning under inductive and transductive settings. Specifically, TMR mainly consists of two components. (1) The topology-aware inductive representation captures information from the directed relations of unseen entities, and aggregates query-related topology features in an attentive manner to generate the fine-grained entity-independent features. (2) After completing multi-modal feature fusion, the relation-augment adaptive RL conducts multi-hop reasoning by eliminating manual rewards and dynamically adding actions. Finally, we construct new MKG datasets with different scales for inductive reasoning evaluation. Experimental results demonstrate that TMP outperforms state-of-the-art MKGR methods under both inductive and transductive settings.
Recent legislation of the "right to be forgotten" has led to the interest in machine unlearning, where the learned models are endowed with the function to forget information about specific training instances as if they have never existed in the training set. Previous work mainly focuses on computer vision scenarios and largely ignores the essentials of unlearning in NLP field, where text data contains more explicit and sensitive personal information than images. In this paper, we propose a general unlearning framework called KGA to induce forgetfulness. Different from previous work that tries to recover gradients or forces models to perform close to one specific distribution, KGA maintains distribution differences (i.e., knowledge gap). This relaxes the distribution assumption. Furthermore, we first apply the unlearning method to various NLP tasks (i.e., classification, translation, response generation) and propose several unlearning evaluation metrics with pertinence. Experiments on large-scale datasets show that KGA yields comprehensive improvements over baselines, where extensive analyses further validate the effectiveness of KGA and provide insight into unlearning for NLP tasks.
Traditional recommender systems estimate user preference on items purely based on historical interaction records, thus failing to capture fine-grained yet dynamic user interests and letting users receive recommendation only passively. Recent conversational recommender systems (CRSs) tackle those limitations by enabling recommender systems to interact with the user to obtain her/his current preference through a sequence of clarifying questions. Despite the progress achieved in CRSs, existing solutions are far from satisfaction in the following two aspects: 1) current CRSs usually require each user to answer a quantity of clarifying questions before reaching the final recommendation, which harms the user experience; 2) there is a semantic gap between the learned representations of explicitly mentioned attributes and items. To address these drawbacks, we introduce the knowledge graph (KG) as the auxiliary information for comprehending and reasoning a user's preference, and propose a new CRS framework, namely Knowledge Enhanced Conversational Reasoning (KECR) system. As a user can reflect her/his preference via both attribute- and item-level expressions, KECR closes the semantic gap between two levels by embedding the structured knowledge in the KG. Meanwhile, KECR utilizes the connectivity within the KG to conduct explicit reasoning of the user demand, making the model less dependent on the user's feedback to clarifying questions. KECR can find a prominent reasoning chain to make the recommendation explainable and more rationale, as well as smoothen the conversation process, leading to better user experience and conversational recommendation accuracy. Extensive experiments on two real-world datasets demonstrate our approach's superiority over state-of-the-art baselines in both automatic evaluations and human judgments.
Latent factor models are the most popular backbones for today's recommender systems owing to their prominent performance. Latent factor models represent users and items as real-valued embedding vectors for pairwise similarity computation, and all embeddings are traditionally restricted to a uniform size that is relatively large (e.g., 256-dimensional). With the exponentially expanding user base and item catalog in contemporary e-commerce, this design is admittedly becoming memory-inefficient. To facilitate lightweight recommendation, reinforcement learning (RL) has recently opened up opportunities for identifying varying embedding sizes for different users/items. However, challenged by search efficiency and learning an optimal RL policy, existing RL-based methods are restricted to highly discrete, predefined embedding size choices. This leads to a largely overlooked potential of introducing finer granularity into embedding sizes to obtain better recommendation effectiveness under a given memory budget. In this paper, we propose continuous input embedding size search (CIESS), a novel RL-based method that operates on a continuous search space with arbitrary embedding sizes to choose from. In CIESS, we further present an innovative random walk-based exploration strategy to allow the RL policy to efficiently explore more candidate embedding sizes and converge to a better decision. CIESS is also model-agnostic and hence generalizable to a variety of latent factor RSs, whilst experiments on two real-world datasets have shown state-of-the-art performance of CIESS under different memory budgets when paired with three popular recommendation models.
Knowledge graphs (KGs) have become important auxiliary information for helping recommender systems obtain a good understanding of user preferences. Despite recent advances in KG-based recommender systems, existing methods are prone to suboptimal performance due to the following two drawbacks: 1) current KG-based methods over-emphasize the heterogeneous structural information within a KG and overlook the underlying semantics of its connections, hindering the recommender from distilling the explicit user preferences; and 2) the inherent incompleteness of a KG (i.e., missing facts, relations and entities) will deteriorate the information extracted from KG and weaken the representation learning of recommender systems. To tackle the aforementioned problems, we investigate the potential of jointly incorporating the structural and semantic information within a KG to model user preferences in finer granularity. A new framework for KG-based recommender systems, namely \textit{K}nowledge \textit{I}nfomax \textit{R}ecommender \textit{S}ystem with \textit{C}ontrastive \textit{L}earning (KIRS-CL) is proposed in this paper. Distinct from previous KG-based approaches, KIRS-CL utilizes structural and connectivity information with high-quality item embeddings learned by encoding KG triples with a pre-trained language model. These well-trained entity representations enable KIRS-CL to find the item to recommend via the preference connection between the user and the item. Additionally, to improve the generalizability of our framework, we introduce a contrastive warm-up learning strategy, making it capable of dealing with both warm- and cold-start recommendation scenarios. Extensive experiments on two real-world datasets demonstrate remarkable improvements over state-of-the-art baselines.
Temporal knowledge graphs (TKGs) model the temporal evolution of events and have recently attracted increasing attention. Since TKGs are intrinsically incomplete, it is necessary to reason out missing elements. Although existing TKG reasoning methods have the ability to predict missing future events, they fail to generate explicit reasoning paths and lack explainability. As reinforcement learning (RL) for multi-hop reasoning on traditional knowledge graphs starts showing superior explainability and performance in recent advances, it has opened up opportunities for exploring RL techniques on TKG reasoning. However, the performance of RL-based TKG reasoning methods is limited due to: (1) lack of ability to capture temporal evolution and semantic dependence jointly; (2) excessive reliance on manually designed rewards. To overcome these challenges, we propose an adaptive reinforcement learning model based on attention mechanism (DREAM) to predict missing elements in the future. Specifically, the model contains two components: (1) a multi-faceted attention representation learning method that captures semantic dependence and temporal evolution jointly; (2) an adaptive RL framework that conducts multi-hop reasoning by adaptively learning the reward functions. Experimental results demonstrate DREAM outperforms state-of-the-art models on public dataset
As an indispensable personalized service in Location-based Social Networks (LBSNs), the next Point-of-Interest (POI) recommendation aims to help people discover attractive and interesting places. Currently, most POI recommenders are based on the conventional centralized paradigm that heavily relies on the cloud to train the recommendation models with large volumes of collected users' sensitive check-in data. Although a few recent works have explored on-device frameworks for resilient and privacy-preserving POI recommendations, they invariably hold the assumption of model homogeneity for parameters/gradients aggregation and collaboration. However, users' mobile devices in the real world have various hardware configurations (e.g., compute resources), leading to heterogeneous on-device models with different architectures and sizes. In light of this, We propose a novel on-device POI recommendation framework, namely Model-Agnostic Collaborative learning for on-device POI recommendation (MAC), allowing users to customize their own model structures (e.g., dimension \& number of hidden layers). To counteract the sparsity of on-device user data, we propose to pre-select neighbors for collaboration based on physical distances, category-level preferences, and social networks. To assimilate knowledge from the above-selected neighbors in an efficient and secure way, we adopt the knowledge distillation framework with mutual information maximization. Instead of sharing sensitive models/gradients, clients in MAC only share their soft decisions on a preloaded reference dataset. To filter out low-quality neighbors, we propose two sampling strategies, performance-triggered sampling and similarity-based sampling, to speed up the training process and obtain optimal recommenders. In addition, we design two novel approaches to generate more effective reference datasets while protecting users' privacy.