Automatic Speech Recognition (ASR) traditionally assumes known domains, but adding data from a new domain raises concerns about computational inefficiencies linked to retraining models on both existing and new domains. Fine-tuning solely on new domain risks Catastrophic Forgetting (CF). To address this, Lifelong Learning (LLL) algorithms have been proposed for ASR. Prior research has explored techniques such as Elastic Weight Consolidation, Knowledge Distillation, and Replay, all of which necessitate either additional parameters or access to prior domain data. We propose Sequential Model Editing as a novel method to continually learn new domains in ASR systems. Different than previous methods, our approach does not necessitate access to prior datasets or the introduction of extra parameters. Our study demonstrates up to 15% Word Error Rate Reduction (WERR) over fine-tuning baseline, and superior efficiency over other LLL techniques on CommonVoice English multi-accent dataset.

Large Language Models (LLMs) excel at providing information acquired during pretraining on large-scale corpora and following instructions through user prompts. This study investigates whether the quality of LLM responses varies depending on the demographic profile of users. Considering English as the global lingua franca, along with the diversity of its dialects among speakers of different native languages, we explore whether non-native English speakers receive lower-quality or even factually incorrect responses from LLMs more frequently. Our results show that performance discrepancies occur when LLMs are prompted by native versus non-native English speakers and persist when comparing native speakers from Western countries with others. Additionally, we find a strong anchoring effect when the model recognizes or is made aware of the user's nativeness, which further degrades the response quality when interacting with non-native speakers. Our analysis is based on a newly collected dataset with over 12,000 unique annotations from 124 annotators, including information on their native language and English proficiency.

In the Minecraft Collaborative Building Task, two players collaborate: an Architect (A) provides instructions to a Builder (B) to assemble a specified structure using 3D blocks. In this work, we investigate the use of large language models (LLMs) to predict the sequence of actions taken by the Builder. Leveraging LLMs' in-context learning abilities, we use few-shot prompting techniques, that significantly improve performance over baseline methods. Additionally, we present a detailed analysis of the gaps in performance for future work

Test-time augmentation (TTA) is a well-known technique employed during the testing phase of computer vision tasks. It involves aggregating multiple augmented versions of input data. Combining predictions using a simple average formulation is a common and straightforward approach after performing TTA. This paper introduces a novel framework for optimizing TTA, called BayTTA (Bayesian-based TTA), which is based on Bayesian Model Averaging (BMA). First, we generate a model list associated with different variations of the input data created through TTA. Then, we use BMA to combine model predictions weighted by their respective posterior probabilities. Such an approach allows one to take into account model uncertainty, and thus to enhance the predictive performance of the related machine learning or deep learning model. We evaluate the performance of BayTTA on various public data, including three medical image datasets comprising skin cancer, breast cancer, and chest X-ray images and two well-known gene editing datasets, CRISPOR and GUIDE-seq. Our experimental results indicate that BayTTA can be effectively integrated into state-of-the-art deep learning models used in medical image analysis as well as into some popular pre-trained CNN models such as VGG-16, MobileNetV2, DenseNet201, ResNet152V2, and InceptionRes-NetV2, leading to the enhancement in their accuracy and robustness performance.

The performance of a large language model (LLM) depends heavily on the quality and size of its pretraining dataset. However, the pretraining datasets for state-of-the-art open LLMs like Llama 3 and Mixtral are not publicly available and very little is known about how they were created. In this work, we introduce FineWeb, a 15-trillion token dataset derived from 96 Common Crawl snapshots that produces better-performing LLMs than other open pretraining datasets. To advance the understanding of how best to curate high-quality pretraining datasets, we carefully document and ablate all of the design choices used in FineWeb, including in-depth investigations of deduplication and filtering strategies. In addition, we introduce FineWeb-Edu, a 1.3-trillion token collection of educational text filtered from FineWeb. LLMs pretrained on FineWeb-Edu exhibit dramatically better performance on knowledge- and reasoning-intensive benchmarks like MMLU and ARC. Along with our datasets, we publicly release our data curation codebase and all of the models trained during our ablation experiments.

Bug localization refers to the identification of source code files which is in a programming language and also responsible for the unexpected behavior of software using the bug report, which is a natural language. As bug localization is labor-intensive, bug localization models are employed to assist software developers. Due to the domain difference between source code files and bug reports, modern bug-localization systems, based on deep learning models, rely heavily on embedding techniques that project bug reports and source code files into a shared vector space. The creation of an embedding involves several design choices, but the impact of these choices on the quality of embedding and the performance of bug localization models remains unexplained in current research. To address this gap, our study evaluated 14 distinct embedding models to gain insights into the effects of various design choices. Subsequently, we developed bug localization models utilizing these embedding models to assess the influence of these choices on the performance of the localization models. Our findings indicate that the pre-training strategies significantly affect the quality of the embedding. Moreover, we discovered that the familiarity of the embedding models with the data has a notable impact on the bug localization model's performance. Notably, when the training and testing data are collected from different projects, the performance of the bug localization models exhibits substantial fluctuations.

The Off-Policy Evaluation (OPE) problem consists of evaluating the performance of counterfactual policies with data collected by another one. This problem is of utmost importance for various application domains, e.g., recommendation systems, medical treatments, and many others. To solve the OPE problem, we resort to estimators, which aim to estimate in the most accurate way possible the performance that the counterfactual policies would have had if they were deployed in place of the logging policy. In the literature, several estimators have been developed, all with different characteristics and theoretical guarantees. Therefore, there is no dominant estimator, and each estimator may be the best one for different OPE problems, depending on the characteristics of the dataset at hand. While the selection of the estimator is a crucial choice for an accurate OPE, this problem has been widely overlooked in the literature. We propose an automated data-driven OPE estimator selection method based on machine learning. In particular, the core idea we propose in this paper is to create several synthetic OPE tasks and use a machine learning model trained to predict the best estimator for those synthetic tasks. We empirically show how our method is able to generalize to unseen tasks and make a better estimator selection compared to a baseline method on several real-world datasets, with a computational cost significantly lower than the one of the baseline.

Over the past few decades, electroencephalography (EEG) monitoring has become a pivotal tool for diagnosing neurological disorders, particularly for detecting seizures. Epilepsy, one of the most prevalent neurological diseases worldwide, affects approximately the 1 \% of the population. These patients face significant risks, underscoring the need for reliable, continuous seizure monitoring in daily life. Most of the techniques discussed in the literature rely on supervised Machine Learning (ML) methods. However, the challenge of accurately labeling variations in epileptic EEG waveforms complicates the use of these approaches. Additionally, the rarity of ictal events introduces an high imbalancing within the data, which could lead to poor prediction performance in supervised learning approaches. Instead, a semi-supervised approach allows to train the model only on data not containing seizures, thus avoiding the issues related to the data imbalancing. This work proposes a semi-supervised approach for detecting epileptic seizures from EEG data, utilizing a novel Deep Learning-based method called SincVAE. This proposal incorporates the learning of an ad-hoc array of bandpass filter as a first layer of a Variational Autoencoder (VAE), potentially eliminating the preprocessing stage where informative band frequencies are identified and isolated. Results indicate that SincVAE improves seizure detection in EEG data and is capable of identifying early seizures during the preictal stage as well as monitoring patients throughout the postictal stage.

This work is motivated by two key trends. On one hand, large language models (LLMs) have shown remarkable versatility in various generative tasks such as writing, drawing, and question answering, significantly reducing the time required for many routine tasks. On the other hand, researchers, whose work is not only time-consuming but also highly expertise-demanding, face increasing challenges as they have to spend more time reading, writing, and reviewing papers. This raises the question: how can LLMs potentially assist researchers in alleviating their heavy workload? This study focuses on the topic of LLMs assist NLP Researchers, particularly examining the effectiveness of LLM in assisting paper (meta-)reviewing and its recognizability. To address this, we constructed the ReviewCritique dataset, which includes two types of information: (i) NLP papers (initial submissions rather than camera-ready) with both human-written and LLM-generated reviews, and (ii) each review comes with "deficiency" labels and corresponding explanations for individual segments, annotated by experts. Using ReviewCritique, this study explores two threads of research questions: (i) "LLMs as Reviewers", how do reviews generated by LLMs compare with those written by humans in terms of quality and distinguishability? (ii) "LLMs as Metareviewers", how effectively can LLMs identify potential issues, such as Deficient or unprofessional review segments, within individual paper reviews? To our knowledge, this is the first work to provide such a comprehensive analysis.

How novel are texts generated by language models (LMs) relative to their training corpora? In this work, we investigate the extent to which modern LMs generate $n$-grams from their training data, evaluating both (i) the probability LMs assign to complete training $n$-grams and (ii) $n$-novelty, the proportion of $n$-grams generated by an LM that did not appear in the training data (for arbitrarily large $n$). To enable arbitrary-length $n$-gram search over a corpus in constant time, we develop Rusty-DAWG, a novel search tool inspired by indexing of genomic data. We compare the novelty of LM-generated text to human-written text and explore factors that affect generation novelty, focusing on the Pythia models. We find that, for $n > 4$, LM-generated text is less novel than human-written text, though it is more novel for smaller $n$. Larger LMs and more constrained decoding strategies both decrease novelty. Finally, we show that LMs complete $n$-grams with lower loss if they are more frequent in the training data. Overall, our results reveal factors influencing the novelty of LM-generated text, and we release Rusty-DAWG to facilitate further pretraining data research.