Many natural language processing (NLP) tasks are naturally imbalanced, as some target categories occur much more frequently than others in the real world. In such scenarios, current NLP models still tend to perform poorly on less frequent classes. Addressing class imbalance in NLP is an active research topic, yet, finding a good approach for a particular task and imbalance scenario is difficult. With this survey, the first overview on class imbalance in deep-learning based NLP, we provide guidance for NLP researchers and practitioners dealing with imbalanced data. We first discuss various types of controlled and real-world class imbalance. Our survey then covers approaches that have been explicitly proposed for class-imbalanced NLP tasks or, originating in the computer vision community, have been evaluated on them. We organize the methods by whether they are based on sampling, data augmentation, choice of loss function, staged learning, or model design. Finally, we discuss open problems such as dealing with multi-label scenarios, and propose systematic benchmarking and reporting in order to move forward on this problem as a community.

The rise of graph representation learning as the primary solution for many different network science tasks led to a surge of interest in the fairness of this family of methods. Link prediction, in particular, has a substantial social impact. However, link prediction algorithms tend to increase the segregation in social networks by disfavoring the links between individuals in specific demographic groups. This paper proposes a novel way to enforce fairness on graph neural networks with a fine-tuning strategy. We Drop the unfair Edges and, simultaneously, we Adapt the model's parameters to those modifications, DEA in short. We introduce two covariance-based constraints designed explicitly for the link prediction task. We use these constraints to guide the optimization process responsible for learning the new "fair" adjacency matrix. One novelty of DEA is that we can use a discrete yet learnable adjacency matrix in our fine-tuning. We demonstrate the effectiveness of our approach on five real-world datasets and show that we can improve both the accuracy and the fairness of the link prediction tasks. In addition, we present an in-depth ablation study demonstrating that our training algorithm for the adjacency matrix can be used to improve link prediction performances during training. Finally, we compute the relevance of each component of our framework to show that the combination of both the constraints and the training of the adjacency matrix leads to optimal performances.

An increasing number of data science and machine learning problems rely on computation with tensors, which better capture the multi-way relationships and interactions of data than matrices. When tapping into this critical advantage, a key challenge is to develop computationally efficient and provably correct algorithms for extracting useful information from tensor data that are simultaneously robust to corruptions and ill-conditioning. This paper tackles tensor robust principal component analysis (RPCA), which aims to recover a low-rank tensor from its observations contaminated by sparse corruptions, under the Tucker decomposition. To minimize the computation and memory footprints, we propose to directly recover the low-dimensional tensor factors -- starting from a tailored spectral initialization -- via scaled gradient descent (ScaledGD), coupled with an iteration-varying thresholding operation to adaptively remove the impact of corruptions. Theoretically, we establish that the proposed algorithm converges linearly to the true low-rank tensor at a constant rate that is independent with its condition number, as long as the level of corruptions is not too large. Empirically, we demonstrate that the proposed algorithm achieves better and more scalable performance than state-of-the-art matrix and tensor RPCA algorithms through synthetic experiments and real-world applications.

Transfer learning has recently become the dominant paradigm of machine learning. Pre-trained models fine-tuned for downstream tasks achieve better performance with fewer labelled examples. Nonetheless, it remains unclear how to develop models that specialise towards multiple tasks without incurring negative interference and that generalise systematically to non-identically distributed tasks. Modular deep learning has emerged as a promising solution to these challenges. In this framework, units of computation are often implemented as autonomous parameter-efficient modules. Information is conditionally routed to a subset of modules and subsequently aggregated. These properties enable positive transfer and systematic generalisation by separating computation from routing and updating modules locally. We offer a survey of modular architectures, providing a unified view over several threads of research that evolved independently in the scientific literature. Moreover, we explore various additional purposes of modularity, including scaling language models, causal inference, programme induction, and planning in reinforcement learning. Finally, we report various concrete applications where modularity has been successfully deployed such as cross-lingual and cross-modal knowledge transfer. Related talks and projects to this survey, are available at https://www.modulardeeplearning.com/.

More than one hundred benchmarks have been developed to test the commonsense knowledge and commonsense reasoning abilities of artificial intelligence (AI) systems. However, these benchmarks are often flawed and many aspects of common sense remain untested. Consequently, we do not currently have any reliable way of measuring to what extent existing AI systems have achieved these abilities. This paper surveys the development and uses of AI commonsense benchmarks. We discuss the nature of common sense; the role of common sense in AI; the goals served by constructing commonsense benchmarks; and desirable features of commonsense benchmarks. We analyze the common flaws in benchmarks, and we argue that it is worthwhile to invest the work needed ensure that benchmark examples are consistently high quality. We survey the various methods of constructing commonsense benchmarks. We enumerate 139 commonsense benchmarks that have been developed: 102 text-based, 18 image-based, 12 video based, and 7 simulated physical environments. We discuss the gaps in the existing benchmarks and aspects of commonsense reasoning that are not addressed in any existing benchmark. We conclude with a number of recommendations for future development of commonsense AI benchmarks.

Multi-domain recommender systems benefit from cross-domain representation learning and positive knowledge transfer. Both can be achieved by introducing a specific modeling of input data (i.e. disjoint history) or trying dedicated training regimes. At the same time, treating domains as separate input sources becomes a limitation as it does not capture the interplay that naturally exists between domains. In this work, we efficiently learn multi-domain representation of sequential users' interactions using graph neural networks. We use temporal intra- and inter-domain interactions as contextual information for our method called MAGRec (short for Multi-domAin Graph-based Recommender). To better capture all relations in a multi-domain setting, we learn two graph-based sequential representations simultaneously: domain-guided for recent user interest, and general for long-term interest. This approach helps to mitigate the negative knowledge transfer problem from multiple domains and improve overall representation. We perform experiments on publicly available datasets in different scenarios where MAGRec consistently outperforms state-of-the-art methods. Furthermore, we provide an ablation study and discuss further extensions of our method.

Global problems, such as pandemics and climate change, require rapid international coordination and diffusion of policy. These phenomena are rare however, with one notable example being the international policy response to the COVID-19 pandemic in early 2020. Here we build an agent-based model of this rapid policy diffusion, where countries constitute the agents and with the principal mechanism for diffusion being peer mimicry. Since it is challenging to predict accurately the policy diffusion curve, we utilize data assimilation, that is an ``on-line'' feed of data to constrain the model against observations. The specific data assimilation algorithm we apply is a particle filter because of its convenient implementation, its ability to handle categorical variables and because the model is not overly computationally expensive, hence a more efficient algorithm is not required. We find that the model alone is able to predict the policy diffusion relatively well with an ensemble of at least 100 simulation runs. The particle filter however improves the fit to the data, reliably so from 500 runs upwards, and increasing filtering frequency results in improved prediction.

The use of AI in healthcare is designed to improve care delivery and augment the decisions of providers to enhance patient outcomes. When deployed in clinical settings, the interaction between providers and AI is a critical component for measuring and understanding the effectiveness of these digital tools on broader health outcomes. Even in cases where AI algorithms have high diagnostic accuracy, healthcare providers often still rely on their experience and sometimes gut feeling to make a final decision. Other times, providers rely unquestioningly on the outputs of the AI models, which leads to a concern about over-reliance on the technology. The purpose of this research was to understand how reliant drug shop dispensers were on AI-powered technologies when determining a differential diagnosis for a presented clinical case vignette. We explored how the drug dispensers responded to technology that is framed as always correct in an attempt to measure whether they begin to rely on it without any critical thought of their own. We found that dispensers relied on the decision made by the AI 25 percent of the time, even when the AI provided no explanation for its decision.

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. Iranian attacks in the waterways of the Middle East and elsewhere in the region "have the attention of everyone" as tensions rise over Tehran's advancing nuclear program, the head of the U.S. Navy's 5th Fleet said Tuesday. Vice Adm. Brad Cooper also told The Associated Press that he's seen a rise in what he described as Iran's "malign activities" in the region over his two years leading the Bahrain-based 5th Fleet. While Cooper pointed to recent seizures of weapons by American and allied forces in the region as a success, he acknowledged that Iran has been able to carry out drone attacks targeting shipping in the Mideast and other assaults in the region.

The world of artificial intelligence (AI) is rapidly advancing, with new discoveries and breakthroughs emerging at an unprecedented pace. For researchers and practitioners in the field, staying current and connected is vital, and attending top AI conferences in 2023 can offer unique opportunities for collaboration, inspiration, and professional growth. From NeurIPS to KDD, these conferences bring together leading experts in machine learning, deep learning, natural language processing, and more. Whether you're an established researcher, an aspiring practitioner, or just passionate about the latest AI developments, these conferences are a must-attend. So join the excitement and start planning your trip to one of these top AI conferences in 2023.