Reliable prediction is an essential requirement for deep neural models that are deployed in open environments, where both covariate and semantic out-of-distribution (OOD) data arise naturally. In practice, to make safe decisions, a reliable model should accept correctly recognized inputs while rejecting both those misclassified covariate-shifted and semantic-shifted examples. Besides, considering the potential existing trade-off between rejecting different failure cases, more convenient, controllable, and flexible failure detection approaches are needed. To meet the above requirements, we propose a simple failure detection framework to unify and facilitate classification with rejection under both covariate and semantic shifts. Our key insight is that by separating and consolidating failure-specific reliability knowledge with low-rank adapters and then integrating them, we can enhance the failure detection ability effectively and flexibly. Extensive experiments demonstrate the superiority of our framework.

As the ability of large language models (LLMs) evolves rapidly, their applications have gradually touched every corner of our daily lives. However, these fast-developing tools raise concerns about the abuse of LLMs. The misuse of LLMs could harm human society in ways such as launching bots on social media, creating fake news and content, and cheating on writing school essays. The overwhelming synthetic data created by the LLMs rather than real humans is also dragging down the efforts to improve the LLMs themselves: the synthetic data pollutes the data pool and should be detected and removed to create a high-quality dataset before training (Radford et al., 2023). Numerous attempts have been made to make the detection possible which can mainly be classified into two categories: post hoc detection that does not modify the language model and the watermarking that changes the output to encode information in the content. Post hoc detection aims to train models that directly label the texts without monitoring the generation process. Although post hoc detections do not require access to modify the output of LLMs, they do make use of statistical features such as the internal activations of the LLMs. For example, when being inspected by another LLM, the statistical properties of machine-generated texts deviate from the human-generated ones in some aspects such as the distributions of token log-likelihoods (Gehrmann et al., 2019; Ippolito et al., 2019; Zellers et al., 2019; Solaiman et al., 2019; Tian, 2023; Mitchell et al., 2023). However, post hoc ways usually rely on the fundamental assumption that machine-generated texts statistically deviate from human-generated texts, which could be challenged in two ways.

With the remarkable generation performance of large language models, ethical and legal concerns about using them have been raised, such as plagiarism and copyright issues. For such concerns, several approaches to watermark and detect LLM-generated text have been proposed very recently. However, we discover that the previous methods fail to function appropriately with code generation tasks because of the syntactic and semantic characteristics of code. Based on \citet{Kirchenbauer2023watermark}, we propose a new watermarking method, Selective WatErmarking via Entropy Thresholding (SWEET), that promotes "green" tokens only at the position with high entropy of the token distribution during generation, thereby preserving the correctness of the generated code. The watermarked code is detected by the statistical test and Z-score based on the entropy information. Our experiments on HumanEval and MBPP show that SWEET significantly improves the Pareto Frontier between the code correctness and watermark detection performance. We also show that notable post-hoc detection methods (e.g. DetectGPT) fail to work well in this task. Finally, we show that setting a reasonable entropy threshold is not much of a challenge. Code is available at https://github.com/hongcheki/sweet-watermark.

As the number of automatic tools based on machine learning (ML) and resting-state electroencephalography (rs-EEG) for Parkinson's disease (PD) detection keeps growing, the assessment of possible exacerbation of health disparities by means of fairness and bias analysis becomes more relevant. Protected attributes, such as gender, play an important role in PD diagnosis development. However, analysis of sub-group populations stemming from different genders is seldom taken into consideration in ML models' development or the performance assessment for PD detection. In this work, we perform a systematic analysis of the detection ability for gender sub-groups in a multi-center setting of a previously developed ML algorithm based on power spectral density (PSD) features of rs-EEG. We find significant differences in the PD detection ability for males and females at testing time (80.5% vs. 63.7% accuracy) and significantly higher activity for a set of parietal and frontal EEG channels and frequency sub-bands for PD and non-PD males that might explain the differences in the PD detection ability for the gender sub-groups.