Deep neural networks (DNNs) deployed in real-world applications can encounter out-of-distribution (OOD) data and adversarial examples. These represent distinct forms of distributional shifts that can significantly impact DNNs' reliability and robustness. Traditionally, research has addressed OOD detection and adversarial robustness as separate challenges. This survey focuses on the intersection of these two areas, examining how the research community has investigated them together. Consequently, we identify two key research directions: robust OOD detection and unified robustness. Robust OOD detection aims to differentiate between in-distribution (ID) data and OOD data, even when they are adversarially manipulated to deceive the OOD detector. Unified robustness seeks a single approach to make DNNs robust against both adversarial attacks and OOD inputs. Accordingly, first, we establish a taxonomy based on the concept of distributional shifts. This framework clarifies how robust OOD detection and unified robustness relate to other research areas addressing distributional shifts, such as OOD detection, open set recognition, and anomaly detection. Subsequently, we review existing work on robust OOD detection and unified robustness. Finally, we highlight the limitations of the existing work and propose promising research directions that explore adversarial and OOD inputs within a unified framework.

In recent years, generative AI (GenAI), like large language models and text-to-image models, has received significant attention across various domains. However, ensuring the responsible generation of content by these models is crucial for their real-world applicability. This raises an interesting question: \textit{What should responsible GenAI generate, and what should it not?} To answer the question, this paper investigates the practical responsible requirements of both textual and visual generative models, outlining five key considerations: generating truthful content, avoiding toxic content, refusing harmful instruction, leaking no training data-related content, and ensuring generated content identifiable. Specifically, we review recent advancements and challenges in addressing these requirements. Besides, we discuss and emphasize the importance of responsible GenAI across healthcare, education, finance, and artificial general intelligence domains. Through a unified perspective on both textual and visual generative models, this paper aims to provide insights into practical safety-related issues and further benefit the community in building responsible GenAI.

The domain of online learning has experienced multifaceted expansion owing to its prevalence in real-life applications. Nonetheless, this progression operates under the assumption that the input feature space of the streaming data remains constant. In this survey paper, we address the topic of online learning in the context of haphazard inputs, explicitly foregoing such an assumption. We discuss, classify, evaluate, and compare the methodologies that are adept at modeling haphazard inputs, additionally providing the corresponding code implementations and their carbon footprint. Moreover, we classify the datasets related to the field of haphazard inputs and introduce evaluation metrics specifically designed for datasets exhibiting imbalance. The code of each methodology can be found at https://github.com/Rohit102497/HaphazardInputsReview

Large language models (LLMs), trained on vast datasets, can carry biases that manifest in various forms, from overt discrimination to implicit stereotypes. One facet of bias is performance disparities in LLMs, often harming underprivileged groups, such as racial minorities. A common approach to quantifying bias is to use template-based bias probes, which explicitly state group membership (e.g. White) and evaluate if the outcome of a task, sentiment analysis for instance, is invariant to the change of group membership (e.g. change White race to Black). This approach is widely used in bias quantification. However, in this work, we find evidence of an unexpectedly overlooked consequence of using template-based probes for LLM bias quantification. We find that in doing so, text examples associated with White ethnicities appear to be classified as exhibiting negative sentiment at elevated rates. We hypothesize that the scenario arises artificially through a mismatch between the pre-training text of LLMs and the templates used to measure bias through reporting bias, unstated norms that imply group membership without explicit statement. Our finding highlights the potential misleading impact of varying group membership through explicit mention in bias quantification

We revisit the likelihood ratio between a pretrained large language model (LLM) and its finetuned variant as a criterion for out-of-distribution (OOD) detection. The intuition behind such a criterion is that, the pretrained LLM has the prior knowledge about OOD data due to its large amount of training data, and once finetuned with the in-distribution data, the LLM has sufficient knowledge to distinguish their difference. Leveraging the power of LLMs, we show that, for the first time, the likelihood ratio can serve as an effective OOD detector. Moreover, we apply the proposed LLM-based likelihood ratio to detect OOD questions in question-answering (QA) systems, which can be used to improve the performance of specialized LLMs for general questions. Given that likelihood can be easily obtained by the loss functions within contemporary neural network frameworks, it is straightforward to implement this approach in practice. Since both the pretrained LLMs and its various finetuned models are available, our proposed criterion can be effortlessly incorporated for OOD detection without the need for further training. We conduct comprehensive evaluation across on multiple settings, including far OOD, near OOD, spam detection, and QA scenarios, to demonstrate the effectiveness of the method.

In this paper, we develop a functional differentiability approach for solving statistical optimal allocation problems. We first derive Hadamard differentiability of the value function through a detailed analysis of the general properties of the sorting operator. Central to our framework are the concept of Hausdorff measure and the area and coarea integration formulas from geometric measure theory. Building on our Hadamard differentiability results, we demonstrate how the functional delta method can be used to directly derive the asymptotic properties of the value function process for binary constrained optimal allocation problems, as well as the two-step ROC curve estimator. Moreover, leveraging profound insights from geometric functional analysis on convex and local Lipschitz functionals, we obtain additional generic Fr\'echet differentiability results for the value functions of optimal allocation problems. These compelling findings motivate us to study carefully the first order approximation of the optimal social welfare. In this paper, we then present a double / debiased estimator for the value functions. Importantly, the conditions outlined in the Hadamard differentiability section validate the margin assumption from the statistical classification literature employing plug-in methods that justifies a faster convergence rate.

Breast cancer has rapidly increased in prevalence in recent years, making it one of the leading causes of mortality worldwide. Among all cancers, it is by far the most common. Diagnosing this illness manually requires significant time and expertise. Since detecting breast cancer is a time-consuming process, preventing its further spread can be aided by creating machine-based forecasts. Machine learning and Explainable AI are crucial in classification as they not only provide accurate predictions but also offer insights into how the model arrives at its decisions, aiding in the understanding and trustworthiness of the classification results. In this study, we evaluate and compare the classification accuracy, precision, recall, and F-1 scores of five different machine learning methods using a primary dataset (500 patients from Dhaka Medical College Hospital). Five different supervised machine learning techniques, including decision tree, random forest, logistic regression, naive bayes, and XGBoost, have been used to achieve optimal results on our dataset. Additionally, this study applied SHAP analysis to the XGBoost model to interpret the model's predictions and understand the impact of each feature on the model's output. We compared the accuracy with which several algorithms classified the data, as well as contrasted with other literature in this field. After final evaluation, this study found that XGBoost achieved the best model accuracy, which is 97%.

Automatic open-domain dialogue evaluation has attracted increasing attention. Trainable evaluation metrics are commonly trained with true positive and randomly selected negative responses, resulting in a tendency for them to assign a higher score to the responses that share higher content similarity with a given context. However, adversarial negative responses possess high content similarity with the contexts whilst being semantically different. Therefore, existing evaluation metrics are not robust enough to evaluate such responses, resulting in low correlations with human judgments. While recent studies have shown some efficacy in utilizing Large Language Models (LLMs) for open-domain dialogue evaluation, they still encounter challenges in effectively handling adversarial negative examples. In this paper, we propose a simple yet effective framework for open-domain dialogue evaluation, which combines domain-specific language models (SLMs) with LLMs. The SLMs can explicitly incorporate Abstract Meaning Representation (AMR) graph information of the dialogue through a gating mechanism for enhanced semantic representation learning. The evaluation result of SLMs and AMR graph information are plugged into the prompt of LLM, for the enhanced in-context learning performance. Experimental results on open-domain dialogue evaluation tasks demonstrate the superiority of our method compared to a wide range of state-of-the-art baselines, especially in discriminating adversarial negative responses.

Adverse drug reactions considerably impact patient outcomes and healthcare costs in cancer therapy. Using artificial intelligence to predict adverse drug reactions in real time could revolutionize oncology treatment. This study aims to assess the performance of artificial intelligence models in predicting adverse drug reactions in patients with cancer. This is the first systematic review and meta-analysis. Scopus, PubMed, IEEE Xplore, and ACM Digital Library databases were searched for studies in English, French, and Arabic from January 1, 2018, to August 20, 2023. The inclusion criteria were: (1) peer-reviewed research articles; (2) use of artificial intelligence algorithms (machine learning, deep learning, knowledge graphs); (3) study aimed to predict adverse drug reactions (cardiotoxicity, neutropenia, nephrotoxicity, hepatotoxicity); (4) study was on cancer patients. The data were extracted and evaluated by three reviewers for study quality. Of the 332 screened articles, 17 studies (5%) involving 93,248 oncology patients from 17 countries were included in the systematic review, of which ten studies synthesized the meta-analysis. A random-effects model was created to pool the sensitivity, specificity, and AUC of the included studies. The pooled results were 0.82 (95% CI:0.69, 0.9), 0.84 (95% CI:0.75, 0.9), and 0.83 (95% CI:0.77, 0.87) for sensitivity, specificity, and AUC, respectively, of ADR predictive models. Biomarkers proved their effectiveness in predicting ADRs, yet they were adopted by only half of the reviewed studies. The use of AI in cancer treatment shows great potential, with models demonstrating high specificity and sensitivity in predicting ADRs. However, standardized research and multicenter studies are needed to improve the quality of evidence. AI can enhance cancer patient care by bridging the gap between data-driven insights and clinical expertise.

Off-policy Evaluation (OPE) methods are a crucial tool for evaluating policies in high-stakes domains such as healthcare, where exploration is often infeasible, unethical, or expensive. However, the extent to which such methods can be trusted under adversarial threats to data quality is largely unexplored. In this work, we make the first attempt at investigating the sensitivity of OPE methods to marginal adversarial perturbations to the data. We design a generic data poisoning attack framework leveraging influence functions from robust statistics to carefully construct perturbations that maximize error in the policy value estimates. We carry out extensive experimentation with multiple healthcare and control datasets. Our results demonstrate that many existing OPE methods are highly prone to generating value estimates with large errors when subject to data poisoning attacks, even for small adversarial perturbations. These findings question the reliability of policy values derived using OPE methods and motivate the need for developing OPE methods that are statistically robust to train-time data poisoning attacks.