Offline Evaluation for Reinforcement Learning-based Recommendation: A Critical Issue and Some Alternatives

In this paper, we argue that the paradigm commonly adopted for offline evaluation of sequential recommender systems is unsuitable for evaluating reinforcement learning-based recommenders. We find that most of the existing offline evaluation practices for reinforcement learning-based recommendation are based on a next-item prediction protocol, and detail three shortcomings of such an evaluation protocol. Notably, it cannot reflect the potential benefits that reinforcement learning (RL) is expected to bring while it hides critical deficiencies of certain offline RL agents. Our suggestions for alternative ways to evaluate RL-based recommender systems aim to shed light on the existing possibilities and inspire future research on reliable evaluation protocols.

Contrastive Learning Reduces Hallucination in Conversations

Pre-trained language models (LMs) store knowledge in their parameters and can generate informative responses when used in conversational systems. However, LMs suffer from the problem of "hallucination:" they may generate plausible-looking statements that are irrelevant or factually incorrect. To address this problem, we propose a contrastive learning scheme, named MixCL. A novel mixed contrastive objective is proposed to explicitly optimize the implicit knowledge elicitation process of LMs, and thus reduce their hallucination in conversations. We also examine negative sampling strategies of retrieved hard negatives and model-generated negatives. We conduct experiments on Wizard-of-Wikipedia, a public, open-domain knowledge-grounded dialogue benchmark, and assess the effectiveness of MixCL. MixCL effectively reduces the hallucination of LMs in conversations and achieves the highest performance among LM-based dialogue agents in terms of relevancy and factuality. We show that MixCL achieves comparable performance to state-of-the-art KB-based approaches while enjoying notable advantages in terms of efficiency and scalability.

Feature-Level Debiased Natural Language Understanding

Natural language understanding (NLU) models often rely on dataset biases rather than intended task-relevant features to achieve high performance on specific datasets. As a result, these models perform poorly on datasets outside the training distribution. Some recent studies address this issue by reducing the weights of biased samples during the training process. However, these methods still encode biased latent features in representations and neglect the dynamic nature of bias, which hinders model prediction. We propose an NLU debiasing method, named debiasing contrastive learning (DCT), to simultaneously alleviate the above problems based on contrastive learning. We devise a debiasing, positive sampling strategy to mitigate biased latent features by selecting the least similar biased positive samples. We also propose a dynamic negative sampling strategy to capture the dynamic influence of biases by employing a bias-only model to dynamically select the most similar biased negative samples. We conduct experiments on three NLU benchmark datasets. Experimental results show that DCT outperforms state-of-the-art baselines on out-of-distribution datasets while maintaining in-distribution performance. We also verify that DCT can reduce biased latent features from the model's representation.

RADio -- Rank-Aware Divergence Metrics to Measure Normative Diversity in News Recommendations

In traditional recommender system literature, diversity is often seen as the opposite of similarity, and typically defined as the distance between identified topics, categories or word models. However, this is not expressive of the social science's interpretation of diversity, which accounts for a news organization's norms and values and which we here refer to as normative diversity. We introduce RADio, a versatile metrics framework to evaluate recommendations according to these normative goals. RADio introduces a rank-aware Jensen Shannon (JS) divergence. This combination accounts for (i) a user's decreasing propensity to observe items further down a list and (ii) full distributional shifts as opposed to point estimates. We evaluate RADio's ability to reflect five normative concepts in news recommendations on the Microsoft News Dataset and six (neural) recommendation algorithms, with the help of our metadata enrichment pipeline. We find that RADio provides insightful estimates that can potentially be used to inform news recommender system design.

Certified Robustness to Word Substitution Ranking Attack for Neural Ranking Models

Neural ranking models (NRMs) have achieved promising results in information retrieval. NRMs have also been shown to be vulnerable to adversarial examples. A typical Word Substitution Ranking Attack (WSRA) against NRMs was proposed recently, in which an attacker promotes a target document in rankings by adding human-imperceptible perturbations to its text. This raises concerns when deploying NRMs in real-world applications. Therefore, it is important to develop techniques that defend against such attacks for NRMs. In empirical defenses adversarial examples are found during training and used to augment the training set. However, such methods offer no theoretical guarantee on the models' robustness and may eventually be broken by other sophisticated WSRAs. To escape this arms race, rigorous and provable certified defense methods for NRMs are needed. To this end, we first define the \textit{Certified Top-$K$ Robustness} for ranking models since users mainly care about the top ranked results in real-world scenarios. A ranking model is said to be Certified Top-$K$ Robust on a ranked list when it is guaranteed to keep documents that are out of the top $K$ away from the top $K$ under any attack. Then, we introduce a Certified Defense method, named CertDR, to achieve certified top-$K$ robustness against WSRA, based on the idea of randomized smoothing. Specifically, we first construct a smoothed ranker by applying random word substitutions on the documents, and then leverage the ranking property jointly with the statistical property of the ensemble to provably certify top-$K$ robustness. Extensive experiments on two representative web search datasets demonstrate that CertDR can significantly outperform state-of-the-art empirical defense methods for ranking models.

Intersection of Parallels as an Early Stopping Criterion

A common way to avoid overfitting in supervised learning is early stopping, where a held-out set is used for iterative evaluation during training to find a sweet spot in the number of training steps that gives maximum generalization. However, such a method requires a disjoint validation set, thus part of the labeled data from the training set is usually left out for this purpose, which is not ideal when training data is scarce. Furthermore, when the training labels are noisy, the performance of the model over a validation set may not be an accurate proxy for generalization. In this paper, we propose a method to spot an early stopping point in the training iterations without the need for a validation set. We first show that in the overparameterized regime the randomly initialized weights of a linear model converge to the same direction during training. Using this result, we propose to train two parallel instances of a linear model, initialized with different random seeds, and use their intersection as a signal to detect overfitting. In order to detect intersection, we use the cosine distance between the weights of the parallel models during training iterations. Noticing that the final layer of a NN is a linear map of pre-last layer activations to output logits, we build on our criterion for linear models and propose an extension to multi-layer networks, using the new notion of counterfactual weights. We conduct experiments on two areas that early stopping has noticeable impact on preventing overfitting of a NN: (i) learning from noisy labels; and (ii) learning to rank in IR. Our experiments on four widely used datasets confirm the effectiveness of our method for generalization. For a wide range of learning rates, our method, called Cosine-Distance Criterion (CDC), leads to better generalization on average than all the methods that we compare against in almost all of the tested cases.

Towards the Use of Saliency Maps for Explaining Low-Quality Electrocardiograms to End Users

When using medical images for diagnosis, either by clinicians or artificial intelligence (AI) systems, it is important that the images are of high quality. When an image is of low quality, the medical exam that produced the image often needs to be redone. In telemedicine, a common problem is that the quality issue is only flagged once the patient has left the clinic, meaning they must return in order to have the exam redone. This can be especially difficult for people living in remote regions, who make up a substantial portion of the patients at Portal Telemedicina, a digital healthcare organization based in Brazil. In this paper, we report on ongoing work regarding (i) the development of an AI system for flagging and explaining low-quality medical images in real-time, (ii) an interview study to understand the explanation needs of stakeholders using the AI system at OurCompany, and, (iii) a longitudinal user study design to examine the effect of including explanations on the workflow of the technicians in our clinics. To the best of our knowledge, this would be the first longitudinal study on evaluating the effects of XAI methods on end-users -- stakeholders that use AI systems but do not have AI-specific expertise. We welcome feedback and suggestions on our experimental setup.

Debiasing Learning for Membership Inference Attacks Against Recommender Systems

Learned recommender systems may inadvertently leak information about their training data, leading to privacy violations. We investigate privacy threats faced by recommender systems through the lens of membership inference. In such attacks, an adversary aims to infer whether a user's data is used to train the target recommender. To achieve this, previous work has used a shadow recommender to derive training data for the attack model, and then predicts the membership by calculating difference vectors between users' historical interactions and recommended items. State-of-the-art methods face two challenging problems: (1) training data for the attack model is biased due to the gap between shadow and target recommenders, and (2) hidden states in recommenders are not observational, resulting in inaccurate estimations of difference vectors. To address the above limitations, we propose a Debiasing Learning for Membership Inference Attacks against recommender systems (DL-MIA) framework that has four main components: (1) a difference vector generator, (2) a disentangled encoder, (3) a weight estimator, and (4) an attack model. To mitigate the gap between recommenders, a variational auto-encoder (VAE) based disentangled encoder is devised to identify recommender invariant and specific features. To reduce the estimation bias, we design a weight estimator, assigning a truth-level score for each difference vector to indicate estimation accuracy. We evaluate DL-MIA against both general recommenders and sequential recommenders on three real-world datasets. Experimental results show that DL-MIA effectively alleviates training and estimation biases simultaneously, and achieves state-of-the-art attack performance.

GLDQN: Explicitly Parameterized Quantile Reinforcement Learning for Waste Reduction

We study the problem of restocking a grocery store's inventory with perishable items over time, from a distributional point of view. The objective is to maximize sales while minimizing waste, with uncertainty about the actual consumption by costumers. This problem is of a high relevance today, given the growing demand for food and the impact of food waste on the environment, the economy, and purchasing power. We frame inventory restocking as a new reinforcement learning task that exhibits stochastic behavior conditioned on the agent's actions, making the environment partially observable. We introduce a new reinforcement learning environment based on real grocery store data and expert knowledge. This environment is highly stochastic, and presents a unique challenge for reinforcement learning practitioners. We show that uncertainty about the future behavior of the environment is not handled well by classical supply chain algorithms, and that distributional approaches are a good way to account for the uncertainty. We also present GLDQN, a new distributional reinforcement learning algorithm that learns a generalized lambda distribution over the reward space. We show that GLDQN outperforms other distributional reinforcement learning approaches in our partially observable environments, in both overall reward and generated waste.