Prior studies show that adopting the annotation diversity shaped by different backgrounds and life experiences and incorporating them into the model learning, i.e. multi-perspective approach, contribute to the development of more responsible models. Thus, in this paper we propose a new framework for designing and further evaluating perspective-aware models on stance detection task,in which multiple annotators assign stances based on a controversial topic. We also share a new dataset established through obtaining both human and LLM annotations. Results show that the multi-perspective approach yields better classification performance (higher F1-scores), outperforming the traditional approaches that use a single ground-truth, while displaying lower model confidence scores, probably due to the high level of subjectivity of the stance detection task.

In the recent past, a popular way of evaluating natural language understanding (NLU), was to consider a model's ability to perform natural language inference (NLI) tasks. In this paper, we investigate if NLI tasks, that are rarely used for LLM evaluation, can still be informative for evaluating LLMs. Focusing on five different NLI benchmarks across six models of different scales, we investigate if they are able to discriminate models of different size and quality and how their accuracies develop during training. Furthermore, we investigate the extent to which the softmax distributions of models align with human distributions in cases where statements are ambiguous or vague. Overall, our results paint a positive picture for the NLI tasks: we find that they are able to discriminate well between models at various stages of training, yet are not (all) saturated. Furthermore, we find that while the similarity of model distributions with human label distributions increases with scale, it is still much higher than the similarity between two populations of humans, making it a potentially interesting statistic to consider.

Subjective NLP tasks usually rely on human annotations provided by multiple annotators, whose judgments may vary due to their diverse backgrounds and life experiences. Traditional methods often aggregate multiple annotations into a single ground truth, disregarding the diversity in perspectives that arises from annotator disagreement. In this preliminary study, we examine the effect of including multiple annotations on model accuracy in classification. Our methodology investigates the performance of perspective-aware classification models in stance detection task and further inspects if annotator disagreement affects the model confidence. The results show that multi-perspective approach yields better classification performance outperforming the baseline which uses the single label. This entails that designing more inclusive perspective-aware AI models is not only an essential first step in implementing responsible and ethical AI, but it can also achieve superior results than using the traditional approaches.

It is common practice in text classification to only use one majority label for model training even if a dataset has been annotated by multiple annotators. Doing so can remove valuable nuances and diverse perspectives inherent in the annotators' assessments. This paper proposes and compares three different strategies to leverage annotator disagreement for text classification: a probability-based multi-label method, an ensemble system, and instruction tuning. All three approaches are evaluated on the tasks of hate speech and abusive conversation detection, which inherently entail a high degree of subjectivity. Moreover, to evaluate the effectiveness of embracing annotation disagreements for model training, we conduct an online survey that compares the performance of the multi-label model against a baseline model, which is trained with the majority label. The results show that in hate speech detection, the multi-label method outperforms the other two approaches, while in abusive conversation detection, instruction tuning achieves the best performance. The results of the survey also show that the outputs from the multi-label models are considered a better representation of the texts than the single-label model.

Federated learning provides a privacy-preserving manner to collaboratively train models on data distributed over multiple local clients via the coordination of a global server. In this paper, we focus on label distribution skew in federated learning, where due to the different user behavior of the client, label distributions between different clients are significantly different. When faced with such cases, most existing methods will lead to a suboptimal optimization due to the inadequate utilization of label distribution information in clients. Inspired by this, we propose a label-masking distillation approach termed FedLMD to facilitate federated learning via perceiving the various label distributions of each client. We classify the labels into majority and minority labels based on the number of examples per class during training. The client model learns the knowledge of majority labels from local data. The process of distillation masks out the predictions of majority labels from the global model, so that it can focus more on preserving the minority label knowledge of the client. A series of experiments show that the proposed approach can achieve state-of-the-art performance in various cases. Moreover, considering the limited resources of the clients, we propose a variant FedLMD-Tf that does not require an additional teacher, which outperforms previous lightweight approaches without increasing computational costs. Our code is available at https://github.com/wnma3mz/FedLMD.

The paper proposes a novel problem in multi-class Multiple-Instance Learning (MIL) called Learning from the Majority Label (LML). In LML, the majority class of instances in a bag is assigned as the bag's label. LML aims to classify instances using bag-level majority classes. This problem is valuable in various applications. Existing MIL methods are unsuitable for LML due to aggregating confidences, which may lead to inconsistency between the bag-level label and the label obtained by counting the number of instances for each class. This may lead to incorrect instance-level classification. We propose a counting network trained to produce the bag-level majority labels estimated by counting the number of instances for each class. This led to the consistency of the majority class between the network outputs and one obtained by counting the number of instances. Experimental results show that our counting network outperforms conventional MIL methods on four datasets The code is publicly available at https://github.com/Shiku-Kaito/Counting-Network-for-Learning-from-Majority-Label.

Researchers recently found out that sometimes language models achieve high accuracy on benchmark data set, but they can not generalize very well with even little changes to the original data set. This is sometimes due to data artifacts, model is learning the spurious correlation between tokens and labels, instead of the semantics and logic. In this work, we analyzed SNLI data and visualized such spurious correlations. We proposed an adaptive up-sampling algorithm to correct the data artifacts, which is simple and effective, and does not need human edits or annotation. We did an experiment applying the algorithm to fix the data artifacts in SNLI data and the model trained with corrected data performed significantly better than the model trained with raw SNLI data, overall, as well as on the subset we corrected.

MultiLabel Learning (MLL) [1] is one of the most common machine learning tasks today. It is based on the idea that each data sample is associated with a certain subset of labels. The full set of labels can be large, in many cases even having more labels than input features. As a result, it is common for some labels to occur in only a few samples, while others occur much more frequently. The label imbalance [2] in MLL is almost always present, and it is a serious obstacle to training good classifiers. Class imbalance is a very well-known problem in traditional learning tasks such as binary and multiclass classification. Hundreds of articles [3, 4, 5], conference papers [6] and books [7] have been devoted to studying it and proposing possible solutions. The most popular are data resampling, cost-sensitive learning and mixtures of these approaches [8, 9]. However, imbalanced learning in the MLL field presents some specific aspects that make it more difficult to deal with this problem.

We give the first algorithm that maintains an approximate decision tree over an arbitrary sequence of insertions and deletions of labeled examples, with strong guarantees on the worst-case running time per update request. For instance, we show how to maintain a decision tree where every vertex has Gini gain within an additive $\alpha$ of the optimum by performing $O\Big(\frac{d\,(\log n)^4}{\alpha^3}\Big)$ elementary operations per update, where $d$ is the number of features and $n$ the maximum size of the active set (the net result of the update requests). We give similar bounds for the information gain and the variance gain. In fact, all these bounds are corollaries of a more general result, stated in terms of decision rules -- functions that, given a set $S$ of labeled examples, decide whether to split $S$ or predict a label. Decision rules give a unified view of greedy decision tree algorithms regardless of the example and label domains, and lead to a general notion of $\epsilon$-approximate decision trees that, for natural decision rules such as those used by ID3 or C4.5, implies the gain approximation guarantees above. The heart of our work provides a deterministic algorithm that, given any decision rule and any $\epsilon > 0$, maintains an $\epsilon$-approximate tree using $O\!\left(\frac{d\, f(n)}{n} \operatorname{poly}\frac{h}{\epsilon}\right)$ operations per update, where $f(n)$ is the complexity of evaluating the rule over a set of $n$ examples and $h$ is the maximum height of the maintained tree.

Despite the subjective nature of many NLP tasks, most NLU evaluations have focused on using the majority label with presumably high agreement as the ground truth. Less attention has been paid to the distribution of human opinions. We collect ChaosNLI, a dataset with a total of 464,500 annotations to study Collective HumAn OpinionS in oft-used NLI evaluation sets. This dataset is created by collecting 100 annotations per example for 3,113 examples in SNLI and MNLI and 1,532 examples in Abductive-NLI. Analysis reveals that: (1) high human disagreement exists in a noticeable amount of examples in these datasets; (2) the state-of-the-art models lack the ability to recover the distribution over human labels; (3) models achieve near-perfect accuracy on the subset of data with a high level of human agreement, whereas they can barely beat a random guess on the data with low levels of human agreement, which compose most of the common errors made by state-of-the-art models on the evaluation sets. This questions the validity of improving model performance on old metrics for the low-agreement part of evaluation datasets. Hence, we argue for a detailed examination of human agreement in future data collection efforts, and evaluating model outputs against the distribution over collective human opinions. The ChaosNLI dataset and experimental scripts are available at https://github.com/easonnie/ChaosNLI