Search

Search query suggestions affect users' interactions with search engines, which then influences the information they encounter. Thus, bias in search query suggestions can lead to exposure to biased search results and can impact opinion formation. This is especially critical in the political domain. Detecting and quantifying bias in web search engines is difficult due to its topic dependency, complexity, and subjectivity. The lack of context and phrasality of query suggestions emphasizes this problem. In a multi-step approach, we combine the benefits of large language models, pairwise comparison, and Elo-based scoring to identify and quantify bias in English search query suggestions. We apply our approach to the U.S. political news domain and compare bias in Google and Bing.

Pre-trained Language Models (PLMs) encode various facts about the world at their pre-training phase as they are trained to predict the next or missing word in a sentence. There has a been an interest in quantifying and improving the amount of facts that can be extracted from PLMs, as they have been envisioned to act as soft knowledge bases, which can be queried in natural language. Different approaches exist to enhance fact retrieval from PLM. Recent work shows that the hidden states of PLMs can be leveraged to determine the truthfulness of the PLMs' inputs. Leveraging this finding to improve factual knowledge retrieval remains unexplored. In this work, we investigate the use of a helper model to improve fact retrieval. The helper model assesses the truthfulness of an input based on the corresponding hidden states representations from the PLMs. We evaluate this approach on several masked PLMs and show that it enhances fact retrieval by up to 33\%. Our findings highlight the potential of hidden states representations from PLMs in improving their factual knowledge retrieval.

Archives of libraries contain many materials, which have not yet been made available to the public. The prioritization of which content to provide and especially how to design effective access paths depend on potential users' needs. As a case study we interviewed researchers working on topics related to one German philosopher to map out their information interaction workflow. Additionally, we deeply analyze study participants' requirements for a digital library system. Moreover, we discuss how existing methods may meet their requirements and which implications these methods may have in a practical digital library setting, e.g., computational costs and hallucinations. In brief, this paper contributes the findings of our digital humanities case study resulting in system requirements., Comment: Technical Report of our accepted JCDL 2024 Poster, 6 pages

In-context knowledge editing (IKE) enables efficient modification of large language model (LLM) outputs without parameter changes and at zero-cost. However, it can be misused to manipulate responses opaquely, e.g., insert misinformation or offensive content. Such malicious interventions could be incorporated into high-level wrapped APIs where the final input prompt is not shown to end-users. To address this issue, we investigate the detection and reversal of IKE-edits. First, we demonstrate that IKE-edits can be detected with high accuracy (F1 > 80\%) using only the top-10 output probabilities of the next token, even in a black-box setting, e.g. proprietary LLMs with limited output information. Further, we introduce the novel task of reversing IKE-edits using specially tuned reversal tokens. We explore using both continuous and discrete reversal tokens, achieving over 80\% accuracy in recovering original, unedited outputs across multiple LLMs. Our continuous reversal tokens prove particularly effective, with minimal impact on unedited prompts. Through analysis of output distributions, attention patterns, and token rankings, we provide insights into IKE's effects on LLMs and how reversal tokens mitigate them. This work represents a significant step towards enhancing LLM resilience against potential misuse of in-context editing, improving their transparency and trustworthiness.

This paper is a report of the Workshop on Simulations for Information Access (Sim4IA) workshop at SIGIR 2024. The workshop had two keynotes, a panel discussion, nine lightning talks, and two breakout sessions. Key takeaways were user simulation's importance in academia and industry, the possible bridging of online and offline evaluation, and the issues of organizing a companion shared task around user simulations for information access. We report on how we organized the workshop, provide a brief overview of what happened at the workshop, and summarize the main topics and findings of the workshop and future work., Comment: Preprint of a SIGIR Forum submission for Vol. 58 No. 2 - December 2024

Reproducibility is essential for scientific research. However, in computer vision, achieving consistent results is challenging due to various factors. One influential, yet often unrecognized, factor is CUDA-induced randomness. Despite CUDA's advantages for accelerating algorithm execution on GPUs, if not controlled, its behavior across multiple executions remains non-deterministic. While reproducibility issues in ML being researched, the implications of CUDA-induced randomness in application are yet to be understood. Our investigation focuses on this randomness across one standard benchmark dataset and two real-world datasets in an isolated environment. Our results show that CUDA-induced randomness can account for differences up to 4.77% in performance scores. We find that managing this variability for reproducibility may entail increased runtime or reduce performance, but that disadvantages are not as significant as reported in previous studies.

The recently discovered altermagnets exhibit collinear magnetic order with zero net magnetization but with unconventional spin-polarized d/g/i-wave band structures, expanding the known paradigms of ferromagnets and antiferromagnets. In addition to novel current-driven electronic transport effects, the unconventional time-reversal symmetry breaking in these systems also makes it possible to obtain a spin response to linearly polarized fields in the optical frequency domain. We show through ab-initio calculations of the prototypical d-wave altermagnet RuO$_2$, with $[C_2\|C_{4z}]$ symmetry combining twofold spin rotation with fourfold lattice rotation, that there is an optical analogue of a spin splitter effect, as the coupling to a linearly polarized exciting laser field makes the d-wave character of the altermagnet directly visible. By magneto-optical measurements on RuO$_2$ films of a few nanometer thickness, we demonstrate the predicted connection between the polarization of an ultrashort pump pulse and the sign and magnitude of a persistent optically excited electronic spin polarization. Our results point to the possibility of exciting and controlling the electronic spin polarization in altermagnets by such ultrashort optical pulses. In addition, the possibility of exciting an electronic spin polarization by linearly polarized optical fields in a compensated system is a unique consequence of the altermagnetic material properties, and our experimental results therefore present an indication for the existence of an altermagnetic phase in ultrathin RuO$_2$ films.

Strategies related to the blockchain concept of Extractable Value (MEV/BEV), such as arbitrage, front- or backrunning create an economic incentive for network nodes to reduce latency. A modified node, that minimizes transaction validation time and neglects to filter invalid transactions in the Ethereum P2P network, introduces a novel attack vector -- Blockchain Amplification Attack. An attacker exploits those modified nodes to amplify an invalid transaction thousands of times, posing a threat to the entire network. To illustrate attack feasibility and practicality in the current mainnet, we 1) identify thousands of similar attacks in the wild, 2) mathematically model propagation mechanism, 3) empirically measure model parameters from our two monitoring nodes, and 4) compare performance with existing Denial-of-Service attacks through local simulation. We show that an attacker can amplify network traffic at modified nodes by a factor of 3,600, and cause economic damages 13,800 times greater than the amount needed to carry out the attack. Despite these risks, aggressive latency reduction may still be profitable enough to justify the existence of modified nodes. To assess this tradeoff, we 1) simulate the transaction validation process in the local network and 2) empirically measure the latency reduction by deploying our modified node in the Ethereum testnet. We conclude with a cost-benefit analysis of skipping validation and provide mitigation strategies against this attack.

We develop a clearance and settlement model for Peer-to-Peer (P2P) energy trading in low-voltage networks. The model enables direct transactions between parties within an open and distributed system and integrates unused capacity while respecting network constraints. We evaluate the model through simulations of different scenarios (normal operating conditions and extreme conditions) for 24-hour time blocks. Our simulations highlight the benefits of our model in a decentralized energy system, notably its ability to deal with high-trade volumes., Comment: To appear in IEEE SmartGridComm'24 - 2024 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm)

When modeling biological responses using Bayesian non-parametric regression, prior information may be available on the shape of the response in the form of non-linear function spaces that define the general shape of the response. To incorporate such information into the analysis, we develop a non-linear functional shrinkage (NLFS) approach that uniformly shrinks the non-parametric fitted function into a non-linear function space while allowing for fits outside of this space when the data suggest alternative shapes. This approach extends existing functional shrinkage approaches into linear subspaces to shrinkage into non-linear function spaces using a Taylor series expansion and corresponding updating of non-linear parameters. We demonstrate this general approach on the Hill model, a popular, biologically motivated model, and show that shrinkage into combined function spaces, i.e., where one has two or more non-linear functions a priori, is straightforward. We demonstrate this approach through synthetic and real data. Computational details on the underlying MCMC sampling are provided with data and analysis available in an online supplement.

Machine learning models are known to learn spurious correlations, i.e., features having strong relations with class labels but no causal relation. Relying on those correlations leads to poor performance in the data groups without these correlations and poor generalization ability. To improve the robustness of machine learning models to spurious correlations, we propose an approach to extract a subnetwork from a fully trained network that does not rely on spurious correlations. The subnetwork is found by the assumption that data points with the same spurious attribute will be close to each other in the representation space when training with ERM, then we employ supervised contrastive loss in a novel way to force models to unlearn the spurious connections. The increase in the worst-group performance of our approach contributes to strengthening the hypothesis that there exists a subnetwork in a fully trained dense network that is responsible for using only invariant features in classification tasks, therefore erasing the influence of spurious features even in the setup of multi spurious attributes and no prior knowledge of attributes labels.

Volumetric neuroimaging examinations like structural Magnetic Resonance Imaging (sMRI) are routinely applied to support the clinical diagnosis of dementia like Alzheimer's Disease (AD). Neuroradiologists examine 3D sMRI to detect and monitor abnormalities in brain morphology due to AD, like global and/or local brain atrophy and shape alteration of characteristic structures. There is a strong research interest in developing diagnostic systems based on Deep Learning (DL) models to analyse sMRI for AD. However, anatomical information extracted from an sMRI examination needs to be interpreted together with patient's age to distinguish AD patterns from the regular alteration due to a normal ageing process. In this context, part-prototype neural networks integrate the computational advantages of DL in an interpretable-by-design architecture and showed promising results in medical imaging applications. We present PIMPNet, the first interpretable multimodal model for 3D images and demographics applied to the binary classification of AD from 3D sMRI and patient's age. Despite age prototypes do not improve predictive performance compared to the single modality model, this lays the foundation for future work in the direction of the model's design and multimodal prototype training process, Comment: Accepted "late-breaking work" at XAI-2024

Domain probe lists--used to determine which URLs to probe for Web censorship--play a critical role in Internet censorship measurement studies. Indeed, the size and accuracy of the domain probe list limits the set of censored pages that can be detected; inaccurate lists can lead to an incomplete view of the censorship landscape or biased results. Previous efforts to generate domain probe lists have been mostly manual or crowdsourced. This approach is time-consuming, prone to errors, and does not scale well to the ever-changing censorship landscape. In this paper, we explore methods for automatically generating probe lists that are both comprehensive and up-to-date for Web censorship measurement. We start from an initial set of 139,957 unique URLs from various existing test lists consisting of pages from a variety of languages to generate new candidate pages. By analyzing content from these URLs (i.e., performing topic and keyword extraction), expanding these topics, and using them as a feed to search engines, our method produces 119,255 new URLs across 35,147 domains. We then test the new candidate pages by attempting to access each URL from servers in eleven different global locations over a span of four months to check for their connectivity and potential signs of censorship. Our measurements reveal that our method discovered over 1,400 domains--not present in the original dataset--we suspect to be blocked. In short, automatically updating probe lists is possible, and can help further automate censorship measurements at scale., Comment: To appear in the Proceedings on Privacy Enhancing Technologies 2024

In the healthcare sector, a consciousness surrounding data privacy and corresponding data protection regulations, as well as heterogeneous and non-harmonized data, pose huge challenges to large-scale data analysis. Moreover, clinical data often involves partially overlapping features, as some observations may be missing due to various reasons, such as differences in procedures, diagnostic tests, or other recorded patient history information across hospitals or institutes. To address the challenges posed by partially overlapping features and incomplete data in clinical datasets, a comprehensive approach is required. Particularly in the domain of medical data, promising outcomes are achieved by federated random forests whenever features align. However, for most standard algorithms, like random forest, it is essential that all data sets have identical parameters. Therefore, in this work the concept of federated random forest is adapted to a setting with partially overlapping features. Moreover, our research assesses the effectiveness of the newly developed federated random forest models for partially overlapping clinical data. For aggregating the federated, globally optimized model, only features available locally at each site can be used. We tackled two issues in federation: (i) the quantity of involved parties, (ii) the varying overlap of features. This evaluation was conducted across three clinical datasets. The federated random forest model even in cases where only a subset of features overlaps consistently demonstrates superior performance compared to its local counterpart. This holds true across various scenarios, including datasets with imbalanced classes. Consequently, federated random forests for partially overlapped data offer a promising solution to transcend barriers in collaborative research and corporate cooperation.

Knowledge editing methods (KEs) can update language models' obsolete or inaccurate knowledge learned from pre-training. However, KEs can be used for malicious applications, e.g., inserting misinformation and toxic content. Knowing whether a generated output is based on edited knowledge or first-hand knowledge from pre-training can increase users' trust in generative models and provide more transparency. Driven by this, we propose a novel task: detecting edited knowledge in language models. Given an edited model and a fact retrieved by a prompt from an edited model, the objective is to classify the knowledge as either unedited (based on the pre-training), or edited (based on subsequent editing). We instantiate the task with four KEs, two LLMs, and two datasets. Additionally, we propose using the hidden state representations and the probability distributions as features for the detection. Our results reveal that, using these features as inputs to a simple AdaBoost classifiers establishes a strong baseline. This classifier requires only a limited amount of data and maintains its performance even in cross-domain settings. Last, we find it more challenging to distinguish edited knowledge from unedited but related knowledge, highlighting the need for further research. Our work lays the groundwork for addressing malicious model editing, which is a critical challenge associated with the strong generative capabilities of LLMs.

Surgery to treat elderly hip fracture patients may cause complications that can lead to early mortality. An early warning system for complications could provoke clinicians to monitor high-risk patients more carefully and address potential complications early, or inform the patient. In this work, we develop a multimodal deep-learning model for post-operative mortality prediction using pre-operative and per-operative data from elderly hip fracture patients. Specifically, we include static patient data, hip and chest images before surgery in pre-operative data, vital signals, and medications administered during surgery in per-operative data. We extract features from image modalities using ResNet and from vital signals using LSTM. Explainable model outcomes are essential for clinical applicability, therefore we compute Shapley values to explain the predictions of our multimodal black box model. We find that i) Shapley values can be used to estimate the relative contribution of each modality both locally and globally, and ii) a modified version of the chain rule can be used to propagate Shapley values through a sequence of models supporting interpretable local explanations. Our findings imply that a multimodal combination of black box models can be explained by propagating Shapley values through the model sequence., Comment: Accepted at World Conference on Explainable Artificial Intelligence; 19 pages, 2 figures, 7 tables

As NLP models become more complex, understanding their decisions becomes more crucial. Counterfactuals (CFs), where minimal changes to inputs flip a model's prediction, offer a way to explain these models. While Large Language Models (LLMs) have shown remarkable performance in NLP tasks, their efficacy in generating high-quality CFs remains uncertain. This work fills this gap by investigating how well LLMs generate CFs for two NLU tasks. We conduct a comprehensive comparison of several common LLMs, and evaluate their CFs, assessing both intrinsic metrics, and the impact of these CFs on data augmentation. Moreover, we analyze differences between human and LLM-generated CFs, providing insights for future research directions. Our results show that LLMs generate fluent CFs, but struggle to keep the induced changes minimal. Generating CFs for Sentiment Analysis (SA) is less challenging than NLI where LLMs show weaknesses in generating CFs that flip the original label. This also reflects on the data augmentation performance, where we observe a large gap between augmenting with human and LLMs CFs. Furthermore, we evaluate LLMs' ability to assess CFs in a mislabelled data setting, and show that they have a strong bias towards agreeing with the provided labels. GPT4 is more robust against this bias and its scores correlate well with automatic metrics. Our findings reveal several limitations and point to potential future work directions., Comment: Accepted to EMNLP Findings 2024

Counterfactual text generation aims to minimally change a text, such that it is classified differently. Judging advancements in method development for counterfactual text generation is hindered by a non-uniform usage of data sets and metrics in related work. We propose CEval, a benchmark for comparing counterfactual text generation methods. CEval unifies counterfactual and text quality metrics, includes common counterfactual datasets with human annotations, standard baselines (MICE, GDBA, CREST) and the open-source language model LLAMA-2. Our experiments found no perfect method for generating counterfactual text. Methods that excel at counterfactual metrics often produce lower-quality text while LLMs with simple prompts generate high-quality text but struggle with counterfactual criteria. By making CEval available as an open-source Python library, we encourage the community to contribute more methods and maintain consistent evaluation in future work.

Recent advances in natural language processing (NLP) can be largely attributed to the advent of pre-trained language models such as BERT and RoBERTa. While these models demonstrate remarkable performance on general datasets, they can struggle in specialized domains such as medicine, where unique domain-specific terminologies, domain-specific abbreviations, and varying document structures are common. This paper explores strategies for adapting these models to domain-specific requirements, primarily through continuous pre-training on domain-specific data. We pre-trained several German medical language models on 2.4B tokens derived from translated public English medical data and 3B tokens of German clinical data. The resulting models were evaluated on various German downstream tasks, including named entity recognition (NER), multi-label classification, and extractive question answering. Our results suggest that models augmented by clinical and translation-based pre-training typically outperform general domain models in medical contexts. We conclude that continuous pre-training has demonstrated the ability to match or even exceed the performance of clinical models trained from scratch. Furthermore, pre-training on clinical data or leveraging translated texts have proven to be reliable methods for domain adaptation in medical NLP tasks., Comment: Accepted at LREC-COLING 2024

Deep learning models have achieved high performance in medical applications, however, their adoption in clinical practice is hindered due to their black-box nature. Self-explainable models, like prototype-based models, can be especially beneficial as they are interpretable by design. However, if the learnt prototypes are of low quality then the prototype-based models are as good as black-box. Having high quality prototypes is a pre-requisite for a truly interpretable model. In this work, we propose a prototype evaluation framework for coherence (PEF-C) for quantitatively evaluating the quality of the prototypes based on domain knowledge. We show the use of PEF-C in the context of breast cancer prediction using mammography. Existing works on prototype-based models on breast cancer prediction using mammography have focused on improving the classification performance of prototype-based models compared to black-box models and have evaluated prototype quality through anecdotal evidence. We are the first to go beyond anecdotal evidence and evaluate the quality of the mammography prototypes systematically using our PEF-C. Specifically, we apply three state-of-the-art prototype-based models, ProtoPNet, BRAIxProtoPNet++ and PIP-Net on mammography images for breast cancer prediction and evaluate these models w.r.t. i) classification performance, and ii) quality of the prototypes, on three public datasets. Our results show that prototype-based models are competitive with black-box models in terms of classification performance, and achieve a higher score in detecting ROIs. However, the quality of the prototypes are not yet sufficient and can be improved in aspects of relevance, purity and learning a variety of prototypes. We call the XAI community to systematically evaluate the quality of the prototypes to check their true usability in high stake decisions and improve such models further., Comment: Accepted at World Conference on Explainable Artificial Intelligence. Cham: Springer Nature Switzerland, 2024; 21 pages, 5 figures, 3 tables

The optical properties of the layered magnet CrSBr are dominated by intralayer excitons: the antiferromagnetic order between the layers makes layer-to-layer charge hopping, and therefore interlayer excitons, spin-forbidden. An external magnetic field, however, continuously drives the magnetic order towards layer-to-layer ferromagnetic, which opens spin-allowed charge-transfer channels between the layers. Here we elaborate how their admixture changes the composition and nature of the excitons, leading to an extension over many layers, and causes a quadratic red-shift with respect to the external magnetic field. We address these effects by ab-initio $GW$-BSE calculations as a function of magnetic field and cast the data into a minimal four-band model to elucidate the interplay between the various interaction and coupling mechanisms. Our findings should be generally valid for antiferromagnetic layered magnets with and without external magnetic fields, and moreover for any couple of layers with different spin directions. Our insights help to systematically address excitons and predict their optical signatures in such systems., Comment: 9 pages, 3 figures

Information from neuroimaging examinations is increasingly used to support diagnoses of dementia, e.g., Alzheimer's disease. While current clinical practice is mainly based on visual inspection and feature engineering, Deep Learning approaches can be used to automate the analysis and to discover new image-biomarkers. Part-prototype neural networks (PP-NN) are an alternative to standard blackbox models, and have shown promising results in general computer vision. PP-NN's base their reasoning on prototypical image regions that are learned fully unsupervised, and combined with a simple-to-understand decision layer. We present PIPNet3D, a PP-NN for volumetric images. We apply PIPNet3D to the clinical diagnosis of Alzheimer's Disease from structural Magnetic Resonance Imaging (sMRI). We assess the quality of prototypes under a systematic evaluation framework, propose new functionally grounded metrics to evaluate brain prototypes and develop an evaluation scheme to assess their coherency with domain experts. Our results show that PIPNet3D is an interpretable, compact model for Alzheimer's diagnosis with its reasoning well aligned to medical domain knowledge. Notably, PIPNet3D achieves the same accuracy as its blackbox counterpart; and removing the remaining clinically irrelevant prototypes from its decision process does not decrease predictive performance., Comment: Accepted at iMIMIC workshop @MICCAI 2024

This paper investigates the thermoelectric properties of solid polymer electrolytes (SPE) containing lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) and sodium bis(trifluoromethanesulfonyl)imide (NaTFSI) salts, along with carbon-based additives of various dimensionalities. Increasing salt concentration leads to higher Seebeck coefficients as a result of the increasing number of free charge carriers and additional, superimposed effects by ion-ion and ion-polymer interactions. NaTFSI-based electrolytes exhibit negative Seebeck coefficients (up to $S = -1.5\,\mathrm{mV\,K^{-1}}$), indicating dominant mobility of $\mathrm{TFSI^-}$ ions. Quasi-one-dimensional carbon nanotubes (CNTs) increase the Seebeck coefficient by a factor of 3. Planar, two-dimensional graphite flakes (GF) moderately enhance it, affecting $\mathrm{Na^+}$ and $\mathrm{TFSI^-}$ ion mobilities and electronic conductivity. Bulky, three-dimensional carbon black (CB) additives induce a unique behavior where the sign of the Seebeck coefficient changes with temperature, presumably due to interaction with $\mathrm{TFSI^-}$ ions within the CB structure. Changes in activation energy and Vogel temperature with salt concentration suggest structural and mechanical modifications in the polymer matrix. The choice of carbon-based additives and salt concentration significantly influences the thermoelectric properties of SPEs thermoelectric properties, providing insights into their potential for thermoelectric applications. Sodium-based electrolytes emerge as promising, sustainable alternatives to lithium-based systems, aligning with sustainable energy research demands.

Biophysical modeling, particularly involving partial differential equations (PDEs), offers significant potential for tailoring disease treatment protocols to individual patients. However, the inverse problem-solving aspect of these models presents a substantial challenge, either due to the high computational requirements of model-based approaches or the limited robustness of deep learning (DL) methods. We propose a novel framework that leverages the unique strengths of both approaches in a synergistic manner. Our method incorporates a DL ensemble for initial parameter estimation, facilitating efficient downstream evolutionary sampling initialized with this DL-based prior. We showcase the effectiveness of integrating a rapid deep-learning algorithm with a high-precision evolution strategy in estimating brain tumor cell concentrations from magnetic resonance images. The DL-Prior plays a pivotal role, significantly constraining the effective sampling-parameter space. This reduction results in a fivefold convergence acceleration and a Dice-score of 95%.

We present a detailed replication study of the BASS framework, an abstractive summarization system based on the notion of Unified Semantic Graphs. Our investigation includes challenges in replicating key components and an ablation study to systematically isolate error sources rooted in replicating novel components. Our findings reveal discrepancies in performance compared to the original work. We highlight the significance of paying careful attention even to reasonably omitted details for replicating advanced frameworks like BASS, and emphasize key practices for writing replicable papers., Comment: This preprint has not undergone peer review or any post-submission improvements or corrections. The Version of Record of this contribution is published in Advances in Information Retrieval, 46th European Conference on Information Retrieval, ECIR 2024. 16 pages, 4 figures

Factual knowledge encoded in Pre-trained Language Models (PLMs) enriches their representations and justifies their use as knowledge bases. Previous work has focused on probing PLMs for factual knowledge by measuring how often they can correctly predict an object entity given a subject and a relation, and improving fact retrieval by optimizing the prompts used for querying PLMs. In this work, we consider a complementary aspect, namely the coherency of factual knowledge in PLMs, i.e., how often can PLMs predict the subject entity given its initial prediction of the object entity. This goes beyond evaluating how much PLMs know, and focuses on the internal state of knowledge inside them. Our results indicate that PLMs have low coherency using manually written, optimized and paraphrased prompts, but including an evidence paragraph leads to substantial improvement. This shows that PLMs fail to model inverse relations and need further enhancements to be able to handle retrieving facts from their parameters in a coherent manner, and to be considered as knowledge bases., Comment: Accepted to EACL Findings 2024