Side Information Research Articles

Generalised Entropy Accumulation

Consider a sequential process in which each step outputs a system Ai\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$A_i$$\\end{document} and updates a side information register E. We prove that if this process satisfies a natural “non-signalling” condition between past outputs and future side information, the min-entropy of the outputs A1,⋯,An\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$A_1, \\dots , A_n$$\\end{document} conditioned on the side information E at the end of the process can be bounded from below by a sum of von Neumann entropies associated with the individual steps. This is a generalisation of the entropy accumulation theorem (EAT) (Dupuis et al. in Commun Math Phys 379: 867–913, 2020), which deals with a more restrictive model of side information: there, past side information cannot be updated in subsequent rounds, and newly generated side information has to satisfy a Markov condition. Due to its more general model of side-information, our generalised EAT can be applied more easily and to a broader range of cryptographic protocols. As examples, we give the first multi-round security proof for blind randomness expansion and a simplified analysis of the E91 QKD protocol. The proof of our generalised EAT relies on a new variant of Uhlmann’s theorem and new chain rules for the Rényi divergence and entropy, which might be of independent interest.

Promoting Machine Abilities of Discovering and Utilizing Knowledge in a Unified Zero-shot Learning Paradigm

Knowledge discovery and utilization are two essential cognitive processes that enable humans to understand the world and extract new insights from their surroundings. These processes have motivated machine learning studies, particularly zero-shot learning (ZS), which seeks to identify unseen concepts through the use of side information. Previous ZS studies primarily focused on utilizing existing knowledge to infer unseen events, yet they overlook the crucial process of knowledge discovery and the integrated modeling of these knowledge-aware processes. In this study, we present a comprehensive ZS learning approach that explores and evaluates the machine's abilities of discovering and utilizing knowledge. More specifically, to emulate human-like knowledge discovery and utilization processes, we propose a novel visual-aware ZS knowledge graph completion task for evaluation, incorporating a traditional ZS image classification task. Technically, we develop a unified ZS learning paradigm named Cognitive Learner (CoLa) to foster the two knowledge-aware abilities. Including a knowledge representation learning (KRL) module and a knowledge adaptation (KA) module, CoLa adapts well to the two specified tasks with the corresponding data. Extensive experiments on large-scale datasets demonstrate CoLa models’ outstanding performance over compared methods in the two ZS tasks, illustrating their superior ability of discovering and utilizing knowledge.

Next-Generation Secure and Reversible Watermarking for Medical Images using Hybrid Radon-Slantlet Transform

This research article proposes a security-enhanced watermarking method for medical images using the Radon and Slantlet transforms. The first step involves transforming the cover image from the spatial domain to the Radon domain. This transformation involves rotation, scaling, and translation through the Radon transform, which alters the locations of concealed bits. As a result, identifying the embedded data poses a considerable challenge. The embedded data cannot be identified without employing the inverse Radon transform. Subsequently, the Radon-transformed image is converted to the frequency domain using the Slantlet transform. Secret bits are incorporated into frequency coefficients during this phase through the pixel pair mapping approach. The final watermarked image is generated by inserting side information into the robust watermarked image. Simulation experiments are carried out to evaluate the imperceptibility of watermarks in medical images, employing metrics such as PSNR and SSIM. The results indicate high imperceptibility, with PSNR values exceeding 45 dB and SSIM values surpassing 0.95 for all tested images. Furthermore, the proposed method's robustness and reversibility are assessed by exposing watermarked images to various attacks. Performance is measured through the BER and NCC. Experimental findings reveal a BER of 0.2 % for the watermarked information, indicating strong resilience against attacks. Additionally, the NCC is determined to be 0.96, highlighting a high level of reversibility in extracting embedded data.

Crossfeat: a transformer-based cross-feature learning model for predicting drug side effect frequency.

Safe drug treatment requires an understanding of the potential side effects. Identifying the frequency of drug side effects can reduce the risks associated with drug use. However, existing computational methods for predicting drug side effect frequencies heavily depend on known drug side effect frequency information. Consequently, these methods face challenges when predicting the side effect frequencies of new drugs. Although a few methods can predict the side effect frequencies of new drugs, they exhibit unreliable performance owing to the exclusion of drug-side effect relationships. This study proposed CrossFeat, a model based on convolutional neural network-transformer architecture with cross-feature learning that can predict the occurrence and frequency of drug side effects for new drugs, even in the absence of information regarding drug-side effect relationships. CrossFeat facilitates the concurrent learning of drugs and side effect information within its transformer architecture. This simultaneous exchange of information enables drugs to learn about their associated side effects, while side effects concurrently acquire information about the respective drugs. Such bidirectional learning allows for the comprehensive integration of drug and side effect knowledge. Our five-fold cross-validation experiments demonstrated that CrossFeat outperforms existing studies in predicting side effect frequencies for new drugs without prior knowledge. Our model offers a promising approach for predicting the drug side effect frequencies, particularly for new drugs where prior information is limited. CrossFeat's superior performance in cross-validation experiments, along with evidence from case studies and ablation experiments, highlights its effectiveness.

FELRec: efficient handling of item cold-start with dynamic representation learning in recommender systems

AbstractRecommender systems suffer from the cold-start problem whenever a new user joins the platform or a new item is added to the catalog. To address item cold-start, we propose to replace the embedding layer in sequential recommenders with a dynamic storage that has no learnable weights and can keep an arbitrary number of representations. In this paper, we present , a large embedding network that refines the existing representations of users and items in a recursive manner, as new information becomes available. In contrast to similar approaches, our model represents new users and items without side information and time-consuming finetuning, instead it runs a single forward pass over a sequence of existing representations. During item cold-start, our method outperforms similar method by 29.50–47.45%. Further, our proposed model generalizes well to previously unseen datasets in zero-shot settings. The source code is publicly available at https://github.com/kweimann/FELRec.

Quantum Representation based Preference Evolution Network for E-commerce recommendation

Quantum theory, originally developed to explain microscopic physical systems, has recently emerged as a novel conceptual and mathematical framework in information science. This paper applies quantum theory to address challenges in E-commerce recommendation, specifically those involving sequential behavior, aiming to mine effective patterns of preference evolution and more accurately predict user interests. Current recommender systems are limited by the sequence length and underutilize side information such as item attributes. To address these issues, we propose a Quantum Representation-based Preference Evolution Network (QRPEN) for E-commerce recommendations. Unlike traditional methods that focus solely on item-ID, our approach integrates a comprehensive set of side information, including both item-ID and attribute data, at each timestamp. We represent item attributes using quantum superposition states and employ density matrices to describe the probability distribution of same-type attributes. These matrices are then transformed into vectors through a quantum measurement-inspired process and fed into a Quasi-RNN model, enabling parallelization and the modeling of longer sequences. This approach effectively captures the dynamic evolution of user preferences. Experiments on public E-commerce datasets demonstrate that QRPEN achieves competitive performance.

Exploiting explicit item–item correlations from knowledge graphs for enhanced sequential recommendation

In recent years, the research of employing knowledge graphs (KGs) in sequential recommendation (SR) has received a lot of attention, since the side information extracted from KGs, especially the information of the correlations between items, indeed helps the SR models achieve better performance. However, many previous KG-based SR models tend to introduce some noise information when learning item embeddings, or insufficiently fuse item–item correlations into their sequential modeling, thus limiting their performance improvements. In this paper, we propose a Distance-Aware Knowledge-based Sequential Recommendation model (DAKSR), which exploits the explicit item–item correlations from KGs to achieve enhanced SR. Specifically, as one critical component in our DAKSR, the distance score matrix (DSM) is first obtained to indicate the correlations between items, and then leveraged in the following three major modules of DAKSR. First, in the Item-Set Embedding layer (ISE) all item embeddings are learned based on DSM, in which the noise information is eliminated effectively. Meanwhile, the Knowledge-Infused Transformer (KIT) incorporates DSM into its attention mechanism to improve the feature extraction. Furthermore, the Knowledge Contrastive Learning module (KCL) also leverages the item–item correlations presented in DSM to generate two credible sequence views, which are used to refine sample representations through a contrastive learning strategy, and thus improve the model’s robustness. Our extensive experiments on three SR benchmarks obviously demonstrate our DAKSR’s superior performance over the state-of-the-art (SOTA) KG-based recommendation models. The implementation of our DAKSR is available at https://github.com/Easonsi/DAKSR for reproducing our experiment results conveniently.

MCAP: Low-Pass GNNs with Matrix Completion for Academic Recommendations

Graph Neural Networks (GNNs) are commonly used and have shown promising performance in recommendation systems. A major branch, Heterogeneous GNNs, models heterogeneous information by leveraging side information for academic paper recommendations. These networks use message passing and high-order propagation to learn representations for users and items. However, existing recommendation methods perform high-order propagation, leading to suboptimal representation learning. To address this issue, this paper proposes a framework called MCAP, which uses relation-aware GNNs and executes low-pass propagation with matrix completion to enhance academic paper recommendations. The framework uses an attention mechanism to learn top- U relationships by constructing a user-user relation graph based on common authors and venues from interacted items. To efficiently and effectively capture semantic-aware similar items, MCAP builds an item-item relation graph by fusing side information of papers using text embedding models (e.g., Mistral) and large language models (e.g., GPT-3.5-Turbo, GLM-4). Finally, the relation-aware user-user and item-item graphs are incorporated into existing GNN-based models to generate representations of users and papers to enhance academic paper recommendations. The effectiveness of the MCAP is validated using four academic datasets, AMiner-PC, AMiner-WeChat, CiteULike, and DBLP, with user-item interactions and side information of papers. Comprehensive experiments show that the MCAP outperforms state-of-the-art models in terms of Recall@5, NDCG@5, and HR@5 with 69.2%, 70.5%, and 77.6% on the AMiner-WeChat dataset. The code for MCAP is available at https://github.com/THUDM/MCAP .

Routing User-Interest Markov Tree for Scalable Personalized Knowledge-Aware Recommendation.

To facilitate more accurate and explainable recommendation, it is crucial to incorporate side information into user-item interactions. Recently, knowledge graph (KG) has attracted much attention in a variety of domains due to its fruitful facts and abundant relations. However, the expanding scale of real-world data graphs poses severe challenges. In general, most existing KG-based algorithms adopt exhaustively hop-by-hop enumeration strategy to search all the possible relational paths, this manner involves extremely high-cost computations and is not scalable with the increase of hop numbers. To overcome these difficulties, in this article, we propose an end-to-end framework Knowledge-tree-routed UseR-Interest Trajectories Network (KURIT-Net). KURIT-Net employs the user-interest Markov trees (UIMTs) to reconfigure a recommendation-based KG, striking a good balance for routing knowledge between short-distance and long-distance relations between entities. Each tree starts from the preferred items for a user and routes the association reasoning paths along the entities in the KG to provide a human-readable explanation for model prediction. KURIT-Net receives entity and relation trajectory embedding (RTE) and fully reflects potential interests of each user by summarizing all reasoning paths in a KG. Besides, we conduct extensive experiments on six public datasets, our KURIT-Net significantly outperforms state-of-the-art approaches and shows its interpretability in recommendation.

Hypergraph contrastive learning for recommendation with side information

PurposeThis paper addresses the limitations of current graph neural network-based recommendation systems, which often neglect the integration of side information and the modeling of complex high-order interactions among nodes. The research motivation stems from the need to enhance recommendation performance by effectively utilizing all available data. We propose a novel method called MSHCN, which leverages hypergraph neural networks to integrate side information and model complex interactions, thereby improving user and item representations.Design/methodology/approachThe MSHCN method employs a hypergraph structure to incorporate various types of side information, including social relationships among users and item attributes, which are essential for enriching user and item representations. The k-means clustering algorithm is utilized to create item-associated hypergraphs, while sentiment analysis on user reviews refines the modeling of user interests. Additionally, hypergraphs are constructed for user-user and item-item interactions based on interaction similarity. MSHCN also incorporates contrastive learning as an auxiliary task to enhance the representation learning process.FindingsExtensive experiments demonstrate that MSHCN significantly outperforms existing recommendation models, particularly in its ability to capture and utilize side information and high-order interactions. This results in superior user and item representations and improved recommendation performance.Originality/valueThe novelty of MSHCN lies in its use of a hypergraph structure to integrate diverse side information and model intricate high-order interactions. The incorporation of contrastive learning as an auxiliary task sets it apart from other hypergraph-based models, providing a significant enhancement in recommendation accuracy.

When latent features meet side information: A preference relation based graph neural network for collaborative filtering

As recommender systems shift from rating-based to interaction-based models, graph neural network-based collaborative filtering models are gaining popularity due to their powerful representation of user-item interactions. However, these models may not produce good item ranking since they focus on explicit preference predictions. Further, these models do not consider side information since they only capture latent feature information of user-item interactions. This study proposes an approach to overcome these two issues by employing preference relation in the graph neural network model for collaborative filtering. Using preference relation ensures the model will generate a good ranking of items. The item side information is integrated into the model through a trainable matrix, which is crucial when the data is highly sparse. The main advantage of this approach is that the model can be generalized to any recommendation scenario where a graph neural network is used for collaborative filtering. Experimental results obtained using the recent RS datasets show that the proposed model outperformed the related baselines.

Correlated source coding with quantum side information at two decoders

Correlated source coding with quantum side information at two decoders

Identifying the psychological effects of nocebo education: results from two pre-registered experiments

Providing treatment side effect information to patients increases the risk of harm due to the nocebo effect. Nocebo education, in which patients learn about nocebo effects, is a novel strategy that can be used across a variety of situations and individuals to decrease unpleasant treatment side effects. It is currently unclear which psychological changes are induced by nocebo education, which is information required to maximize this intervention. Two pre-registered studies investigated the effects of nocebo education on side effect expectations, side effect control beliefs, feelings toward treatments, intentions to avoid or seek side effect information, and perceptions of treatment efficacy. In Study 1 (N = 220), adult participants either watched or did not watch a nocebo education intervention video prior to reading vignettes about receiving a surgical treatment for pain and a medication for pain. Study 2 (N = 252) was similar to Study 1, with the inclusion of a health behavior video control group and participants only reading about a medication treatment for pain. In both experiments, nocebo education reduced global side effect expectations and increased side effect self-efficacy beliefs. Nocebo education also increased intentions to avoid side effect information and decreased intentions to seek more side effect information. Evidence was inconclusive on whether nocebo education changes affective associations with the treatments. The findings demonstrate that nocebo education has a multi-faceted influence with the potential to change patient behavior. The results can be used to improve the management of adverse treatment side effects.

Efficient On-Board Compression for Arbitrary-Shaped Cloud-Covered Remote Sensing Images via Adaptive Filling and Controllable Quantization

Due to the fact that invalid cloud-covered regions in remote sensing images consume a considerable quantity of coding bit rates under the limited satellite-to-ground transmission rate, existing image compression methods suffer from low compression efficiency and poor reconstruction quality, especially in cloud-free regions which are generally regarded as regions of interest (ROIs). Therefore, we propose an efficient on-board compression method for remote sensing images with arbitrary-shaped clouds by leveraging the characteristics of cloudy images. Firstly, we introduce two novel spatial preprocessing strategies, namely, the optimized adaptive filling (OAF) strategy and the controllable quantization (CQ) strategy. Specifically, the OAF strategy fills each cloudy region using the contextual information at its inner and outer edge to completely remove the information of cloudy regions and minimize their coding consumption, which is suitable for images with only thick clouds. The CQ strategy implicitly identifies thin and thick clouds and rationally quantifies the data in cloudy regions to alleviate information loss in thin cloud-covered regions, which can achieve the balance between coding efficiency and reconstructed image quality and is more suitable for images containing thin clouds. Secondly, we develop an efficient coding method for a binary cloud mask to effectively save the bit rate of the side information. Our method provides the flexibility for users to choose the desired preprocessing strategy as needed and can be embedded into existing compression framework such as JPEG2000. Experimental results on the GF-1 dataset show that our method effectively reduces the coding consumption of invalid cloud-covered regions and significantly improve the compression efficiency as well as the quality of decoded images.

Directed Topic Extraction with Side Information for Sustainability Analysis

Directed Topic Extraction with Side Information for Sustainability Analysis

Enhanced side information fusion framework for sequential recommendation

Enhanced side information fusion framework for sequential recommendation

Optimal innovation-based attacks against remote state estimation with side information and historical data

This work investigates the attack strategy design problem of the false data injection attack against the cyber-physical system. Distinct from the relevant results which utilise the current intercepted data or additionally consider side information, a more universal attack model is proposed which combines historical data and side information with the current intercepted data to synergistically deteriorate the system estimation performance. In order to quantify the impact resulting from the proposed attack strategy, the optimisation objective is characterised by deriving the error covariance matrix under the attacks, which becomes more intricate since the proposed attack model introduces more decision variables and coupled terms. Take the stealthiness which is characterised by Kullback-Leibler divergence as the constraints, the problem investigated in this work is equivalently transformed into the constrained multi-variable non-convex optimisation problems, which are not able to be solved directly by the methods in the relevant results. By utilising the Lagrange multiplier method, the structural characteristic of the optimal mean and the optimal covariance which only related to the Lagrange multiplier are derived, such that the optimal distribution of the modified innovation is able to be obtained by a simple search procedure. Following that, the design of the optimal attack strategy is completed by using semi-definite programming to derive the optimal attack matrices. Finally, the simulation examples are given such that the validity of the results is verified.

RSSI-based attacks for identification of BLE devices

To prevent tracking, the Bluetooth Low Energy (BLE) protocol integrates privacy mechanisms such as address randomization. However, as highlighted by previous researches address randomization is not a silver bullet and can be circumvented by exploiting other types of information disclosed by the protocol such as counters or timing. In this work, we propose two novel attack to break address randomization in BLE exploiting side information in the form of Received Signal Strength Indication (RSSI). More precisely, we demonstrate how RSSI measurements, extracted from received BLE advertising packets, can be used to link together the traces emitted by the same device or directly re-identify it despite address randomization. The proposed attacks leverage the distribution of RSSI to create a fingerprint of devices with an empirical evaluation on various scenarios demonstrating their effectiveness. For instance in the static context, in which devices remain at the same position, the proposed approach yields a re-identification accuracy of up to 97%, which can even be boosted to perfect accuracy by increasing the number of receivers controlled by the adversary. We also discuss the factors influencing the success of the attacks and evaluate two possible countermeasures whose effectiveness is limited, highlighting the difficulty in mitigating this threat.

Multi-source fully test-time adaptation

Deep neural networks have significantly advanced various fields. However, these models often encounter difficulties in achieving effective generalization when the distribution of test samples varies from that of the training samples. Recently, some fully test-time adaptation methods have been proposed to adapt the trained model with the unlabeled test samples before prediction to enhance the test performance. Despite achieving remarkable results, these methods only involve one trained model, which could only provide certain side information for the test samples. In real-world scenarios, there could be multiple available trained models that are beneficial to the test samples and are complementary to each other. Consequently, to better utilize these trained models, in this paper, we propose the problem of multi-source fully test-time adaptation to adapt multiple trained models to the test samples. To address this problem, we introduce a simple yet effective method utilizing a weighted aggregation scheme and introduce two unsupervised losses. The former could adaptively assign a higher weight to a more relevant model, while the latter could jointly adapt models with online unlabeled samples. Extensive experiments on three image classification datasets show that the proposed method achieves better results than baseline methods, demonstrating the superiority in adapting to multiple models.

Syntax-Guided Content-Adaptive Transform for Image Compression.

The surge in image data has significantly increased the pressure on storage and transmission, posing new challenges for image compression technology. The structural texture of an image implies its statistical characteristics, which is effective for image encoding and decoding. Consequently, content-adaptive compression methods based on learning can better capture the content attributes of images, thereby enhancing encoding performance. However, learned image compression methods do not comprehensively account for both the global and local correlations among the pixels within an image. Moreover, they are constrained by rate-distortion optimization, which prevents the attainment of a compact representation of image attributes. To address these issues, we propose a syntax-guided content-adaptive transform framework that efficiently captures image attributes and enhances encoding efficiency. Firstly, we propose a syntax-refined side information module that fully leverages syntax and side information to guide the adaptive transformation of image attributes. Moreover, to more thoroughly exploit the global and local correlations in image space, we designed global-local modules, local-global modules, and upsampling/downsampling modules in codecs, further eliminating local and global redundancies. The experimental findings indicate that our proposed syntax-guided content-adaptive image compression model successfully adapts to the diverse complexities of different images, which enhances the efficiency of image compression. Concurrently, the method proposed has demonstrated outstanding performance across three benchmark datasets.