When answering questions, LLMs can convey not only an answer, but a level of confidence about the answer being correct. This includes explicit confidence markers (e.g. giving a numeric score) as well as implicit markers, like an authoritative tone or elaborating with additional knowledge. For LLMs to be trustworthy knowledge sources, the confidence they convey should match their actual expertise; however, most current models tend towards overconfidence. To calibrate both implicit and explicit confidence markers, we introduce a pragmatic, listener-aware finetuning method (LACIE) that models the listener, considering not only whether an answer is right, but whether it will be accepted by a listener. We cast calibration as preference optimization, creating data via a two-agent game, where a speaker model's outputs are judged by a simulated listener. We then finetune three LLMs (Mistral-7B, Llama3-8B, Llama3-70B) with LACIE, and show that the resulting models are better calibrated w.r.t. a simulated listener. Crucially, these trends transfer to human listeners, helping them correctly predict model correctness: we conduct a human evaluation where annotators accept or reject an LLM's answers, finding that training with LACIE results in 47% fewer incorrect answers being accepted while maintaining the same level of acceptance for correct answers. Furthermore, LACIE generalizes to another dataset, resulting in a large increase in truthfulness on TruthfulQA when trained on TriviaQA. Our analysis indicates that LACIE leads to a better confidence separation between correct and incorrect examples. Qualitatively, we find that a LACIE-trained model hedges more and implicitly signals certainty when it is correct by using an authoritative tone or including details. Finally, LACIE finetuning leads to an emergent increase in model abstention (e.g. saying "I don't know") for answers that are likely wrong.

We introduce a fairness-aware dataset for job recommendation in advertising, designed to foster research in algorithmic fairness within real-world scenarios. It was collected and prepared to comply with privacy standards and business confidentiality. An additional challenge is the lack of access to protected user attributes such as gender, for which we propose a solution to obtain a proxy estimate. Despite being anonymized and including a proxy for a sensitive attribute, our dataset preserves predictive power and maintains a realistic and challenging benchmark. This dataset addresses a significant gap in the availability of fairness-focused resources for high-impact domains like advertising -- the actual impact being having access or not to precious employment opportunities, where balancing fairness and utility is a common industrial challenge. We also explore various stages in the advertising process where unfairness can occur and introduce a method to compute a fair utility metric for the job recommendations in online systems case from a biased dataset. Experimental evaluations of bias mitigation techniques on the released dataset demonstrate potential improvements in fairness and the associated trade-offs with utility.

Classification systems are evaluated in a countless number of papers. However, we find that evaluation practice is often nebulous. Frequently, metrics are selected without arguments, and blurry terminology invites misconceptions. For instance, many works use so-called 'macro' metrics to rank systems (e.g., 'macro F1') but do not clearly specify what they would expect from such a `macro' metric. This is problematic, since picking a metric can affect research findings, and thus any clarity in the process should be maximized. Starting from the intuitive concepts of bias and prevalence, we perform an analysis of common evaluation metrics. The analysis helps us understand the metrics' underlying properties, and how they align with expectations as found expressed in papers. Then we reflect on the practical situation in the field, and survey evaluation practice in recent shared tasks. We find that metric selection is often not supported with convincing arguments, an issue that can make a system ranking seem arbitrary. Our work aims at providing overview and guidance for more informed and transparent metric selection, fostering meaningful evaluation.

We applied machine learning to the unmet medical need of rapid and accurate diagnosis and prognosis of acute infections and sepsis in emergency departments. Our solution consists of a Myrna (TM) Instrument and embedded TriVerity (TM) classifiers. The instrument measures abundances of 29 messenger RNAs in patient's blood, subsequently used as features for machine learning. The classifiers convert the input features to an intuitive test report comprising the separate likelihoods of (1) a bacterial infection (2) a viral infection, and (3) severity (need for Intensive Care Unit-level care). In internal validation, the system achieved AUROC = 0.83 on the three-class disease diagnosis (bacterial, viral, or non-infected) and AUROC = 0.77 on binary prognosis of disease severity. The Myrna, TriVerity system was granted breakthrough device designation by the United States Food and Drug Administration (FDA). This engineering manuscript teaches the standard and novel machine learning methods used to translate an academic research concept to a clinical product aimed at improving patient care, and discusses lessons learned.

The task of multi-hop link prediction within knowledge graphs (KGs) stands as a challenge in the field of knowledge graph analysis, as it requires the model to reason through and understand all intermediate connections before making a prediction. In this paper, we introduce the Knowledge Graph Large Language Model (KG-LLM), a novel framework that leverages large language models (LLMs) for knowledge graph tasks. We first convert structured knowledge graph data into natural language and then use these natural language prompts to fine-tune LLMs to enhance multihop link prediction in KGs. By converting the KG to natural language prompts, our framework is designed to learn the latent representations of entities and their interrelations. To show the efficacy of the KG-LLM Framework, we fine-tune three leading LLMs within this framework, including Flan-T5, LLaMa2 and Gemma. Further, we explore the framework's potential to provide LLMs with zero-shot capabilities for handling previously unseen prompts. Experimental results show that KG-LLM significantly improves the models' generalization capabilities, leading to more accurate predictions in unfamiliar scenarios. Our code is available at https://anonymous.4open.science/r/KG-LLM-FED0.

In this paper, we explore the properties of loss curvature with respect to input data in deep neural networks. Curvature of loss with respect to input (termed input loss curvature) is the trace of the Hessian of the loss with respect to the input. We investigate how input loss curvature varies between train and test sets, and its implications for train-test distinguishability. We develop a theoretical framework that derives an upper bound on the train-test distinguishability based on privacy and the size of the training set. This novel insight fuels the development of a new black box membership inference attack utilizing input loss curvature. We validate our theoretical findings through experiments in computer vision classification tasks, demonstrating that input loss curvature surpasses existing methods in membership inference effectiveness. Our analysis highlights how the performance of membership inference attack (MIA) methods varies with the size of the training set, showing that curvature-based MIA outperforms other methods on sufficiently large datasets. This condition is often met by real datasets, as demonstrated by our results on CIFAR10, CIFAR100, and ImageNet. These findings not only advance our understanding of deep neural network behavior but also improve the ability to test privacy-preserving techniques in machine learning.

Context: Machine learning (ML) and deep learning (DL) analyze raw data to extract valuable insights in specific phases. The rise of continuous practices in software projects emphasizes automating Continuous Integration (CI) with these learning-based methods, while the growing adoption of such approaches underscores the need for systematizing knowledge. Objective: Our objective is to comprehensively review and analyze existing literature concerning learning-based methods within the CI domain. We endeavour to identify and analyse various techniques documented in the literature, emphasizing the fundamental attributes of training phases within learning-based solutions in the context of CI. Method: We conducted a Systematic Literature Review (SLR) involving 52 primary studies. Through statistical and thematic analyses, we explored the correlations between CI tasks and the training phases of learning-based methodologies across the selected studies, encompassing a spectrum from data engineering techniques to evaluation metrics. Results: This paper presents an analysis of the automation of CI tasks utilizing learning-based methods. We identify and analyze nine types of data sources, four steps in data preparation, four feature types, nine subsets of data features, five approaches for hyperparameter selection and tuning, and fifteen evaluation metrics. Furthermore, we discuss the latest techniques employed, existing gaps in CI task automation, and the characteristics of the utilized learning-based techniques. Conclusion: This study provides a comprehensive overview of learning-based methods in CI, offering valuable insights for researchers and practitioners developing CI task automation. It also highlights the need for further research to advance these methods in CI.

Deepfake technology is rapidly advancing, posing significant challenges to the detection of manipulated media content. Parallel to that, some adversarial attack techniques have been developed to fool the deepfake detectors and make deepfakes even more difficult to be detected. This paper explores the application of super resolution techniques as a possible adversarial attack in deepfake detection. Through our experiments, we demonstrate that minimal changes made by these methods in the visual appearance of images can have a profound impact on the performance of deepfake detection systems. We propose a novel attack using super resolution as a quick, black-box and effective method to camouflage fake images and/or generate false alarms on pristine images. Our results indicate that the usage of super resolution can significantly impair the accuracy of deepfake detectors, thereby highlighting the vulnerability of such systems to adversarial attacks.

Parameter-efficient fine-tuning (PEFT) for personalizing automatic speech recognition (ASR) has recently shown promise for adapting general population models to atypical speech. However, these approaches assume a priori knowledge of the atypical speech disorder being adapted for -- the diagnosis of which requires expert knowledge that is not always available. Even given this knowledge, data scarcity and high inter/intra-speaker variability further limit the effectiveness of traditional fine-tuning. To circumvent these challenges, we first identify the minimal set of model parameters required for ASR adaptation. Our analysis of each individual parameter's effect on adaptation performance allows us to reduce Word Error Rate (WER) by half while adapting 0.03% of all weights. Alleviating the need for cohort-specific models, we next propose the novel use of a meta-learned hypernetwork to generate highly individualized, utterance-level adaptations on-the-fly for a diverse set of atypical speech characteristics. Evaluating adaptation at the global, cohort and individual-level, we show that hypernetworks generalize better to out-of-distribution speakers, while maintaining an overall relative WER reduction of 75.2% using 0.1% of the full parameter budget.

Differential privacy (DP) is a widely used approach for mitigating privacy risks when training machine learning models on sensitive data. DP mechanisms add noise during training to limit the risk of information leakage. The scale of the added noise is critical, as it determines the trade-off between privacy and utility. The standard practice is to select the noise scale in terms of a privacy budget parameter $\epsilon$. This parameter is in turn interpreted in terms of operational attack risk, such as accuracy, or sensitivity and specificity of inference attacks against the privacy of the data. We demonstrate that this two-step procedure of first calibrating the noise scale to a privacy budget $\epsilon$, and then translating $\epsilon$ to attack risk leads to overly conservative risk assessments and unnecessarily low utility. We propose methods to directly calibrate the noise scale to a desired attack risk level, bypassing the intermediate step of choosing $\epsilon$. For a target attack risk, our approach significantly decreases noise scale, leading to increased utility at the same level of privacy. We empirically demonstrate that calibrating noise to attack sensitivity/specificity, rather than $\epsilon$, when training privacy-preserving ML models substantially improves model accuracy for the same risk level. Our work provides a principled and practical way to improve the utility of privacy-preserving ML without compromising on privacy.