Fine-tuning pre-trained models is a widely employed technique in numerous real-world applications. However, fine-tuning these models on new tasks can lead to unfair outcomes. This is due to the absence of generalization guarantees for fairness properties, regardless of whether the original pre-trained model was developed with fairness considerations. To tackle this issue, we introduce an efficient and robust fine-tuning framework specifically designed to mitigate biases in new tasks. Our empirical analysis shows that the parameters in the pre-trained model that affect predictions for different demographic groups are different, so based on this observation, we employ a transfer learning strategy that neutralizes the importance of these influential weights, determined using Fisher information across demographic groups. Additionally, we integrate this weight importance neutralization strategy with a matrix factorization technique, which provides a low-rank approximation of the weight matrix using fewer parameters, reducing the computational demands. Experiments on multiple pre-trained models and new tasks demonstrate the effectiveness of our method.

Language Models (LMs) such as BERT, have been shown to perform well on the task of identifying Named Entities (NE) in text. A BERT LM is typically used as a classifier to classify individual tokens in the input text, or to classify spans of tokens, as belonging to one of a set of possible NE categories. In this paper, we hypothesise that decoder-only Large Language Models (LLMs) can also be used generatively to extract both the NE, as well as potentially recover the correct surface form of the NE, where any spelling errors that were present in the input text get automatically corrected. We fine-tune two BERT LMs as baselines, as well as eight open-source LLMs, on the task of producing NEs from text that was obtained by applying Optical Character Recognition (OCR) to images of Japanese shop receipts; in this work, we do not attempt to find or evaluate the location of NEs in the text. We show that the best fine-tuned LLM performs as well as, or slightly better than, the best fine-tuned BERT LM, although the differences are not significant. However, the best LLM is also shown to correct OCR errors in some cases, as initially hypothesised.

Self-consistency has emerged as a powerful method for improving the accuracy of short answers generated by large language models. As previously defined, it only concerns the accuracy of a final answer parsed from generated text. In this work, we extend the idea to open response generation, by integrating voting into the decoding method. Each output sentence is selected from among multiple samples, conditioning on the previous selections, based on a simple token overlap score. We compare this "Sample & Select" method to greedy decoding, beam search, nucleus sampling, and the recently introduced hallucination avoiding decoders of DoLA, P-CRR, and S-CRR. We show that Sample & Select improves factuality by a 30% relative margin against these decoders in NLI-based evaluation on the subsets of CNN/DM and XSum used in the FRANK benchmark, while maintaining comparable ROUGE-1 F1 scores against reference summaries. We collect human verifications of the generated summaries, confirming the factual superiority of our method.

Recent work exploring the capabilities of pre-trained large language models (LLMs) has demonstrated their ability to act as general pattern machines by completing complex token sequences representing a wide array of tasks, including time-series prediction and robot control. This paper investigates whether the pattern recognition and sequence modeling capabilities of LLMs can be extended to the domain of knowledge tracing, a critical component in the development of intelligent tutoring systems (ITSs) that tailor educational experiences by predicting learner performance over time. In an empirical evaluation across multiple real-world datasets, we compare two approaches to using LLMs for this task, zero-shot prompting and model fine-tuning, with existing, non-LLM approaches to knowledge tracing. While LLM-based approaches do not achieve state-of-the-art performance, fine-tuned LLMs surpass the performance of naive baseline models and perform on par with standard Bayesian Knowledge Tracing approaches across multiple metrics. These findings suggest that the pattern recognition capabilities of LLMs can be used to model complex learning trajectories, opening a novel avenue for applying LLMs to educational contexts. The paper concludes with a discussion of the implications of these findings for future research, suggesting that further refinements and a deeper understanding of LLMs' predictive mechanisms could lead to enhanced performance in knowledge tracing tasks.

Optimization metrics are crucial for building recommendation systems at scale. However, an effective and efficient metric for practical use remains elusive. While Top-K ranking metrics are the gold standard for optimization, they suffer from significant computational overhead. Alternatively, the more efficient accuracy and AUC metrics often fall short of capturing the true targets of recommendation tasks, leading to suboptimal performance. To overcome this dilemma, we propose a new optimization metric, Lower-Left Partial AUC (LLPAUC), which is computationally efficient like AUC but strongly correlates with Top-K ranking metrics. Compared to AUC, LLPAUC considers only the partial area under the ROC curve in the Lower-Left corner to push the optimization focus on Top-K. We provide theoretical validation of the correlation between LLPAUC and Top-K ranking metrics and demonstrate its robustness to noisy user feedback. We further design an efficient point-wise recommendation loss to maximize LLPAUC and evaluate it on three datasets, validating its effectiveness and robustness.

This paper explores the influence of Italian high school students' proficiency in mathematics and the Italian language on their university enrolment choices, specifically focusing on STEM (Science, Technology, Engineering, and Mathematics) courses. We distinguish between students from scientific and humanistic backgrounds in high school, providing valuable insights into their enrolment preferences. Furthermore, we investigate potential gender differences in response to similar previous educational choices and achievements. The study employs gradient boosting methodology, known for its high predicting performance and ability to capture non-linear relationships within data, and adjusts for variables related to the socio-demographic characteristics of the students and their previous educational achievements. Our analysis reveals significant differences in the enrolment choices based on previous high school achievements. The findings shed light on the complex interplay of academic proficiency, gender, and high school background in shaping students' choices regarding university education, with implications for educational policy and future research endeavours.

Standardized benchmarks drive progress in machine learning. However, with repeated testing, the risk of overfitting grows as algorithms over-exploit benchmark idiosyncrasies. In our work, we seek to mitigate this challenge by compiling ever-expanding large-scale benchmarks called Lifelong Benchmarks. As exemplars of our approach, we create Lifelong-CIFAR10 and Lifelong-ImageNet, containing (for now) 1.69M and 1.98M test samples, respectively. While reducing overfitting, lifelong benchmarks introduce a key challenge: the high cost of evaluating a growing number of models across an ever-expanding sample set. To address this challenge, we also introduce an efficient evaluation framework: Sort \& Search (S&S), which reuses previously evaluated models by leveraging dynamic programming algorithms to selectively rank and sub-select test samples, enabling cost-effective lifelong benchmarking. Extensive empirical evaluations across 31,000 models demonstrate that S&S achieves highly-efficient approximate accuracy measurement, reducing compute cost from 180 GPU days to 5 GPU hours (1000x reduction) on a single A100 GPU, with low approximation error. As such, lifelong benchmarks offer a robust, practical solution to the "benchmark exhaustion" problem.

Deep learning has shown to have great potential in medical applications. In critical domains as such, it is of high interest to have trustworthy algorithms which are able to tell when reliable assessments cannot be guaranteed. Detecting out-of-distribution (OOD) samples is a crucial step towards building a safe classifier. Following a previous study, showing that it is possible to classify breast cancer in point-of-care ultrasound images, this study investigates OOD detection using three different methods: softmax, energy score and deep ensembles. All methods are tested on three different OOD data sets. The results show that the energy score method outperforms the softmax method, performing well on two of the data sets. The ensemble method is the most robust, performing the best at detecting OOD samples for all three OOD data sets.

Rationales in the form of manually annotated input spans usually serve as ground truth when evaluating explainability methods in NLP. They are, however, time-consuming and often biased by the annotation process. In this paper, we debate whether human gaze, in the form of webcam-based eye-tracking recordings, poses a valid alternative when evaluating importance scores. We evaluate the additional information provided by gaze data, such as total reading times, gaze entropy, and decoding accuracy with respect to human rationale annotations. We compare WebQAmGaze, a multilingual dataset for information-seeking QA, with attention and explainability-based importance scores for 4 different multilingual Transformer-based language models (mBERT, distil-mBERT, XLMR, and XLMR-L) and 3 languages (English, Spanish, and German). Our pipeline can easily be applied to other tasks and languages. Our findings suggest that gaze data offers valuable linguistic insights that could be leveraged to infer task difficulty and further show a comparable ranking of explainability methods to that of human rationales.

Knowledge graph representation learning (KGRL) or knowledge graph embedding (KGE) plays a crucial role in AI applications for knowledge construction and information exploration. These models aim to encode entities and relations present in a knowledge graph into a lower-dimensional vector space. During the training process of KGE models, using positive and negative samples becomes essential for discrimination purposes. However, obtaining negative samples directly from existing knowledge graphs poses a challenge, emphasizing the need for effective generation techniques. The quality of these negative samples greatly impacts the accuracy of the learned embeddings, making their generation a critical aspect of KGRL. This comprehensive survey paper systematically reviews various negative sampling (NS) methods and their contributions to the success of KGRL. Their respective advantages and disadvantages are outlined by categorizing existing NS methods into five distinct categories. Moreover, this survey identifies open research questions that serve as potential directions for future investigations. By offering a generalization and alignment of fundamental NS concepts, this survey provides valuable insights for designing effective NS methods in the context of KGRL and serves as a motivating force for further advancements in the field.