Automatic evaluation metrics are crucial to the development of generative systems. In recent years, pre-trained language model (PLM) based metrics, such as BERTScore, have been commonly adopted in various generation tasks. However, it has been demonstrated that PLMs encode a range of stereotypical societal biases, leading to a concern on the fairness of PLMs as metrics. To that end, this work presents the first systematic study on the social bias in PLM-based metrics. We demonstrate that popular PLM-based metrics exhibit significantly higher social bias than traditional metrics on 6 sensitive attributes, namely race, gender, religion, physical appearance, age, and socioeconomic status. In-depth analysis suggests that choosing paradigms (matching, regression, or generation) of the metric has a greater impact on fairness than choosing PLMs. In addition, we develop debiasing adapters that are injected into PLM layers, mitigating bias in PLM-based metrics while retaining high performance for evaluating text generation.

We propose a benchmark to assess the capability of large language models to reason with conventional metaphors. Our benchmark combines the previously isolated topics of metaphor detection and commonsense reasoning into a single task that requires a model to make inferences by accurately selecting between the literal and metaphorical register. We examine the performance of state-of-the-art pre-trained models on binary-choice tasks and find a large discrepancy between the performance of small and very large models, going from chance to near-human level. We also analyse the largest model in a generative setting and find that although human performance is approached, careful multiple-shot prompting is required.

Natural language processing (NLP) models are known to be vulnerable to backdoor attacks, which poses a newly arisen threat to NLP models. Prior online backdoor defense methods for NLP models only focus on the anomalies at either the input or output level, still suffering from fragility to adaptive attacks and high computational cost. In this work, we take the first step to investigate the unconcealment of textual poisoned samples at the intermediate-feature level and propose a feature-based efficient online defense method. Through extensive experiments on existing attacking methods, we find that the poisoned samples are far away from clean samples in the intermediate feature space of a poisoned NLP model. Motivated by this observation, we devise a distance-based anomaly score (DAN) to distinguish poisoned samples from clean samples at the feature level. Experiments on sentiment analysis and offense detection tasks demonstrate the superiority of DAN, as it substantially surpasses existing online defense methods in terms of defending performance and enjoys lower inference costs. Moreover, we show that DAN is also resistant to adaptive attacks based on feature-level regularization. Our code is available at https://github.com/lancopku/DAN.

Current robotic haptic object recognition relies on statistical measures derived from movement dependent interaction signals such as force, vibration or position. Mechanical properties that can be identified from these signals are intrinsic object properties that may yield a more robust object representation. Therefore, this paper proposes an object recognition framework using multiple representative mechanical properties: the coefficient of restitution, stiffness, viscosity and friction coefficient. These mechanical properties are identified in real-time using a dual Kalman filter, then used to classify objects. The proposed framework was tested with a robot identifying 20 objects through haptic exploration. The results demonstrate the technique's effectiveness and efficiency, and that all four mechanical properties are required for best recognition yielding a rate of 98.18 $\pm$ 0.424 %. Clustering with Gaussian mixture models further shows that using these mechanical properties results in superior recognition as compared to using statistical parameters of the interaction signals.

For a microlensing event with multiple lenses, the interpretation of the light curve can be challenging. First When a distant star (called the source) gets sufficiently of all, the computation of the multiple-lens microlensing aligned with a massive foreground object (called light curve can be time-consuming due to the finitesource the lens), the gravitational field of the lens focuses the effect (e.g., Dong et al. 2006; Bozza 2010). This light out of the distant star, thus making the distant star is especially true when the microlens system consists of appear brighter (Einstein 1936; Paczynski 1986). For a three or more objects (e.g., Gaudi et al. 2008; Kuang typical source star inside the Milky Way, one can observe et al. 2021). Additionally, the likelihood landscape of the time evolution of their brightness (i.e., light curves) the high-dimensional parameter space can be so pathological and infer the existence and properties of companion objects that traditional sampling-based methods may to the lens by monitoring the deviations in the light have a hard time searching for the correct solution (or curve from the single lens scenario (e.g., Mao & Paczynski solutions). This remains to be true even when the brute 1991; Gould & Loeb 1992). This so-called gravitational force search on a fine grid that is defined by a subset microlensing technique has been frequently used of model parameters is conducted. As a result, the current to detect exoplanets and stellar binaries and are complementary analysis of multiple-lens microlensing events is still to other techniques (see reviews by Gaudi case-by-case, with each event requiring hundreds of (or 2012 and Zhu & Dong 2021).

In this work, we take the named entity recognition task in the English language as a case study and explore style transfer as a data augmentation method to increase the size and diversity of training data in low-resource scenarios. We propose a new method to effectively transform the text from a high-resource domain to a low-resource domain by changing its style-related attributes to generate synthetic data for training. Moreover, we design a constrained decoding algorithm along with a set of key ingredients for data selection to guarantee the generation of valid and coherent data. Experiments and analysis on five different domain pairs under different data regimes demonstrate that our approach can significantly improve results compared to current state-of-the-art data augmentation methods. Our approach is a practical solution to data scarcity, and we expect it to be applicable to other NLP tasks.

We provide a quantitative and qualitative analysis of self-repetition in the output of neural summarizers. We measure self-repetition as the number of n-grams of length four or longer that appear in multiple outputs of the same system. We analyze the behavior of three popular architectures (BART, T5, and Pegasus), fine-tuned on five datasets. In a regression analysis, we find that the three architectures have different propensities for repeating content across output summaries for inputs, with BART being particularly prone to self-repetition. Fine-tuning on more abstractive data, and on data featuring formulaic language, is associated with a higher rate of self-repetition. In qualitative analysis we find systems produce artefacts such as ads and disclaimers unrelated to the content being summarized, as well as formulaic phrases common in the fine-tuning domain. Our approach to corpus-level analysis of self-repetition may help practitioners clean up training data for summarizers and ultimately support methods for minimizing the amount of self-repetition.

Diffusion models have shown exceptional scaling properties in the image synthesis domain, and initial attempts have shown similar benefits for applying diffusion to unconditional text synthesis. Denoising diffusion models attempt to iteratively refine a sampled noise signal until it resembles a coherent signal (such as an image or written sentence). In this work we aim to see whether the benefits of diffusion models can also be realized for speech recognition. To this end, we propose a new way to perform speech recognition using a diffusion model conditioned on pretrained speech features. Specifically, we propose TransFusion: a transcribing diffusion model which iteratively denoises a random character sequence into coherent text corresponding to the transcript of a conditioning utterance. We demonstrate comparable performance to existing high-performing contrastive models on the LibriSpeech speech recognition benchmark. To the best of our knowledge, we are the first to apply denoising diffusion to speech recognition. We also propose new techniques for effectively sampling and decoding multinomial diffusion models. These are required because traditional methods of sampling from acoustic models are not possible with our new discrete diffusion approach.

Named Entity Recognition and Disambiguation (NERD) systems are foundational for information retrieval, question answering, event detection, and other natural language processing (NLP) applications. We introduce TweetNERD, a dataset of 340K+ Tweets across 2010-2021, for benchmarking NERD systems on Tweets. This is the largest and most temporally diverse open sourced dataset benchmark for NERD on Tweets and can be used to facilitate research in this area. We describe evaluation setup with TweetNERD for three NERD tasks: Named Entity Recognition (NER), Entity Linking with True Spans (EL), and End to End Entity Linking (End2End); and provide performance of existing publicly available methods on specific TweetNERD splits.

We introduce the User-Aware Arabic Gender Rewriter, a user-centric web-based system for Arabic gender rewriting in contexts involving two users. The system takes either Arabic or English sentences as input, and provides users with the ability to specify their desired first and/or second person target genders. The system outputs gender rewritten alternatives of the Arabic input sentences (or their Arabic translations in case of English input) to match the target users' gender preferences.