Interpretable machine learning models offer understandable reasoning behind their decision-making process, though they may not always match the performance of their black-box counterparts. This trade-off between interpretability and model performance has sparked discussions around the deployment of AI, particularly in critical applications where knowing the rationale of decision-making is essential for trust and accountability. In this study, we conduct a comparative analysis of several black-box and interpretable models, focusing on a specific NLP use case that has received limited attention: inferring ratings from reviews. Through this use case, we explore the intricate relationship between the performance and interpretability of different models. We introduce a quantitative score called Composite Interpretability (CI) to help visualize the trade-off between interpretability and performance, particularly in the case of composite models. Our results indicate that, in general, the learning performance improves as interpretability decreases, but this relationship is not strictly monotonic, and there are instances where interpretable models are more advantageous.

Despite their remarkable potential, Large Vision-Language Models (LVLMs) still face challenges with object hallucination, a problem where their generated outputs mistakenly incorporate objects that do not actually exist. Although most works focus on addressing this issue within the language-model backbone, our work shifts the focus to the image input source, investigating how specific image tokens contribute to hallucinations. Our analysis reveals a striking finding: a small subset of image tokens with high attention scores are the primary drivers of object hallucination. By removing these hallucinatory image tokens (only 1.5% of all image tokens), the issue can be effectively mitigated. This finding holds consistently across different models and datasets. Building on this insight, we introduce EAZY, a novel, training-free method that automatically identifies and Eliminates hAllucinations by Zeroing out hallucinatorY image tokens. We utilize EAZY for unsupervised object hallucination detection, achieving 15% improvement compared to previous methods. Additionally, EAZY demonstrates remarkable effectiveness in mitigating hallucinations while preserving model utility and seamlessly adapting to various LVLM architectures.

Recent inductive logic programming (ILP) approaches learn optimal hypotheses. An optimal hypothesis minimises a given cost function on the training data. There are many cost functions, such as minimising training error, textual complexity, or the description length of hypotheses. However, selecting an appropriate cost function remains a key question. To address this gap, we extend a constraint-based ILP system to learn optimal hypotheses for seven standard cost functions. We then empirically compare the generalisation error of optimal hypotheses induced under these standard cost functions. Our results on over 20 domains and 1000 tasks, including game playing, program synthesis, and image reasoning, show that, while no cost function consistently outperforms the others, minimising training error or description length has the best overall performance. Notably, our results indicate that minimising the size of hypotheses does not always reduce generalisation error.

In recent years, there has been increased interest in the design, training, and evaluation of end-to-end autonomous driving (AD) systems. One often overlooked aspect is the uncertainty of planned trajectories predicted by these systems, despite awareness of their own uncertainty being key to achieve safety and robustness. We propose to estimate this uncertainty by adapting loss prediction from the uncertainty quantification literature. To this end, we introduce a novel light-weight module, dubbed CATPlan, that is trained to decode motion and planning embeddings into estimates of the collision loss used to partially supervise end-to-end AD systems. During inference, these estimates are interpreted as collision risk. We evaluate CATPlan on the safety-critical, nerf-based, closed-loop benchmark NeuroNCAP and find that it manages to detect collisions with a $54.8\%$ relative improvement to average precision over a GMM-based baseline in which the predicted trajectory is compared to the forecasted trajectories of other road users. Our findings indicate that the addition of CATPlan can lead to safer end-to-end AD systems and hope that our work will spark increased interest in uncertainty quantification for such systems.

Object detection systems must reliably perceive objects of interest without being overly confident to ensure safe decision-making in dynamic environments. Filtering techniques based on out-of-distribution (OoD) detection are commonly added as an extra safeguard to filter hallucinations caused by overconfidence in novel objects. Nevertheless, evaluating YOLO-family detectors and their filters under existing OoD benchmarks often leads to unsatisfactory performance. This paper studies the underlying reasons for performance bottlenecks and proposes a methodology to improve performance fundamentally. Our first contribution is a calibration of all existing evaluation results: Although images in existing OoD benchmark datasets are claimed not to have objects within in-distribution (ID) classes (i.e., categories defined in the training dataset), around 13% of objects detected by the object detector are actually ID objects. Dually, the ID dataset containing OoD objects can also negatively impact the decision boundary of filters. These ultimately lead to a significantly imprecise performance estimation. Our second contribution is to consider the task of hallucination reduction as a joint pipeline of detectors and filters. By developing a methodology to carefully synthesize an OoD dataset that semantically resembles the objects to be detected, and using the crafted OoD dataset in the fine-tuning of YOLO detectors to suppress the objectness score, we achieve a 88% reduction in overall hallucination error with a combined fine-tuned detection and filtering system on the self-driving benchmark BDD-100K. Our code and dataset are available at: https://gricad-gitlab.univ-grenoble-alpes.fr/dnn-safety/m-hood.

AI Biases as Asymmetries: A Review to Guide Practice Gabriella Waters (CEAMLS, Morgan State University)* Phillip Honenberger (CEAMLS, Morgan State University)* *Equal contribution [Preprint - Nov. 21, 2024] Abstract The understanding of bias in AI is currently undergoing a revolution. Initially understood as errors or flaws, biases are increasingly recognized as integral to AI systems and sometimes preferable to less biased alternatives. In this paper we review the reasons for this changed understanding and provide new guidance on two questions: First, how should we think about and measure biases in AI systems, consistent with the new understanding? Second, what kinds of bias in an AI system should we accept or even amplify, and what kinds should we minimize or eliminate, and why? The key to answering both questions, we argue, is to understand biases as "violations of a symmetry standard" (following Kelly). We distinguish three main types of asymmetry in AI systems - error biases, inequality biases, and process biases - and highlight places in the pipeline of AI development and application where bias of each type is likely to be good, bad, or inevitable. Introduction The understanding of bias in AI is currently undergoing a revolution. Initially perceived as errors or flaws, biases are increasingly recognized as integral to AI systems and sometimes preferable to less biased alternatives. Cognitive psychology and statistics have informed this shift, highlighting the benefits and costs of biases in decision-making processes. Cognitive psychology presents biases as often helpful in making decisions under conditions of uncertainty. Similarly, statistical methods acknowledge biases as often useful and sometimes necessary for making inferences from data. These insights have been instrumental in redefining biases as not inherently negative, but as sometimes essential components that can and should be harnessed to improve AI systems.

The classification of electrocardiogram (ECG) signals is crucial for early detection of arrhythmias and other cardiac conditions. However, despite advances in machine learning, many studies fail to follow standardization protocols, leading to inconsistencies in performance evaluation and real-world applicability. Additionally, hardware constraints essential for practical deployment, such as in pacemakers, Holter monitors, and wearable ECG patches, are often overlooked. Since real-world impact depends on feasibility in resource-constrained devices, ensuring efficient deployment is critical for continuous monitoring. This review systematically analyzes ECG classification studies published between 2017 and 2024, focusing on those adhering to the E3C (Embedded, Clinical, and Comparative Criteria), which include inter-patient paradigm implementation, compliance with Association for the Advancement of Medical Instrumentation (AAMI) recommendations, and model feasibility for embedded systems. While many studies report high accuracy, few properly consider patient-independent partitioning and hardware limitations. We identify state-of-the-art methods meeting E3C criteria and conduct a comparative analysis of accuracy, inference time, energy consumption, and memory usage. Finally, we propose standardized reporting practices to ensure fair comparisons and practical applicability of ECG classification models. By addressing these gaps, this study aims to guide future research toward more robust and clinically viable ECG classification systems.

Deep neural networks are prone to various bias issues, jeopardizing their applications for high-stake decision-making. Existing fairness methods typically offer a fixed accuracy-fairness trade-off, since the weight of the well-trained model is a fixed point (fairness-optimum) in the weight space. Nevertheless, more flexible accuracy-fairness trade-offs at inference time are practically desired since: 1) stakes of the same downstream task can vary for different individuals, and 2) different regions have diverse laws or regularization for fairness. If using the previous fairness methods, we have to train multiple models, each offering a specific level of accuracy-fairness trade-off. This is often computationally expensive, time-consuming, and difficult to deploy, making it less practical for real-world applications. To address this problem, we propose You Only Debias Once (YODO) to achieve in-situ flexible accuracy-fairness trade-offs at inference time, using a single model that trained only once. Instead of pursuing one individual fixed point (fairness-optimum) in the weight space, we aim to find a "line" in the weight space that connects the accuracy-optimum and fairness-optimum points using a single model. Points (models) on this line implement varying levels of accuracy-fairness trade-offs. At inference time, by manually selecting the specific position of the learned "line", our proposed method can achieve arbitrary accuracy-fairness trade-offs for different end-users and scenarios. Experimental results on tabular and image datasets show that YODO achieves flexible trade-offs between model accuracy and fairness, at ultra-low overheads. For example, if we need $100$ levels of trade-off on the \acse dataset, YODO takes $3.53$ seconds while training $100$ fixed models consumes $425$ seconds. The code is available at https://github.com/ahxt/yodo.

Machine learning in materials science faces challenges due to limited experimental data, as generating synthesis data is costly and time-consuming, especially with in-house experiments. Mining data from existing literature introduces issues like mixed data quality, inconsistent formats, and variations in reporting experimental parameters, complicating the creation of consistent features for the learning algorithm. Additionally, combining continuous and discrete features can hinder the learning process with limited data. Here, we propose strategies that utilize large language models (LLMs) to enhance machine learning performance on a limited, heterogeneous dataset of graphene chemical vapor deposition synthesis compiled from existing literature. These strategies include prompting modalities for imputing missing data points and leveraging large language model embeddings to encode the complex nomenclature of substrates reported in chemical vapor deposition experiments. The proposed strategies enhance graphene layer classification using a support vector machine (SVM) model, increasing binary classification accuracy from 39% to 65% and ternary accuracy from 52% to 72%. We compare the performance of the SVM and a GPT-4 model, both trained and fine-tuned on the same data. Our results demonstrate that the numerical classifier, when combined with LLM-driven data enhancements, outperforms the standalone LLM predictor, highlighting that in data-scarce scenarios, improving predictive learning with LLM strategies requires more than simple fine-tuning on datasets. Instead, it necessitates sophisticated approaches for data imputation and feature space homogenization to achieve optimal performance. The proposed strategies emphasize data enhancement techniques, offering a broadly applicable framework for improving machine learning performance on scarce, inhomogeneous datasets.

We identify the slow liquidity drain (SLID) scam, an insidious and highly profitable threat to decentralized finance (DeFi), posing a large-scale, persistent, and growing risk to the ecosystem. Unlike traditional scams such as rug pulls or honeypots (USENIX Sec'19, USENIX Sec'23), SLID gradually siphons funds from liquidity pools over extended periods, making detection significantly more challenging. In this paper, we conducted the first large-scale empirical analysis of 319,166 liquidity pools across six major decentralized exchanges (DEXs) since 2018. We identified 3,117 SLID affected liquidity pools, resulting in cumulative losses of more than US$103 million. We propose a rule-based heuristic and an enhanced machine learning model for early detection. Our machine learning model achieves a detection speed 4.77 times faster than the heuristic while maintaining 95% accuracy. Our study establishes a foundation for protecting DeFi investors at an early stage and promoting transparency in the DeFi ecosystem.