Abstract: Multi-hop Machine reading comprehension is a challenging task with aim of answering a question based on disjoint pieces of information across the different passages. The evaluation metrics and datasets are a vital part of multi-hop MRC because it is not possible to train and evaluate models without them, also, the proposed challenges by datasets often are an important motivation for improving the existing models. Due to increasing attention to this field, it is necessary and worth reviewing them in detail. This study aims to present a comprehensive survey on recent advances in multi-hop MRC evaluation metrics and datasets. In this regard, first, the multi-hop MRC problem definition will be presented, then the evaluation metrics based on their multi-hop aspect will be investigated. Also, 15 multi-hop datasets have been reviewed in detail from 2017 to 2022, and a comprehensive analysis has been prepared at the end. Finally, open issues in this field have been discussed. Keywords: Multi-hop Machine Reading Comprehension, Multi-hop Machine Reading Comprehension Dataset, Natural Language Processing, 1-INTRODUCTION Machine reading comprehension (MRC) is one of the most important and long-standing topics in Natural Language Processing (NLP). MRC provides a way to evaluate an NLP system's capability for natural language understanding. An MRC task, in brief, refers to the ability of a computer to read and understand natural language context and then find the answer to questions about that context. The emergence of large-scale single-document MRC datasets, such as SQuAD (Rajpurkar et al., 2016), CNN/Daily mail (Hermann et al., 2015), has led to increased attention to this topic and different models have been proposed to address the MRC problem, such as (D. However, for many of these datasets, it has been found that models don't need to comprehend and reason to answer a question. For example, Khashabi et al (Khashabi et al., 2016) proved that adversarial perturbation in candidate answers has a negative effect on the performance of the QA systems. Similarly, (Jia & Liang, 2017) showed that adding an adversarial sentence to the SQuAD (Rajpurkar et al., 2016) context will drop the result of many existing models.

This paper presents an algorithm to solve the Soft k-Means problem globally. Unlike Fuzzy c-Means, Soft k-Means (SkM) has a matrix factorization-type objective and has been shown to have a close relation with the popular probability decomposition-type clustering methods, e.g., Left Stochastic Clustering (LSC). Though some work has been done for solving the Soft k-Means problem, they usually use an alternating minimization scheme or the projected gradient descent method, which cannot guarantee global optimality since the non-convexity of SkM. In this paper, we present a sufficient condition for a feasible solution of Soft k-Means problem to be globally optimal and show the output of the proposed algorithm satisfies it. Moreover, for the Soft k-Means problem, we provide interesting discussions on stability, solutions non-uniqueness, and connection with LSC. Then, a new model, named Minimal Volume Soft k-Means (MVSkM), is proposed to address the solutions non-uniqueness issue. Finally, experimental results support our theoretical results.

The people in the world who are hearing impaired face many obstacles in communication and require an interpreter to comprehend what a person is saying. There has been constant scientific research and the existing models lack the ability to make accurate predictions. So we propose a deep learning model trained on ASL i.e. American Sign Language which will take actions in the form of ASL as input and translate it into text. To achieve the translation a Convolution Neural Network model and a transfer learning model based on the VGG16 architecture are used. There has been an improvement in accuracy from 94% of CNN to 98.7% of Transfer Learning, an improvement of 5%. An application with the deep learning model integrated has also been built.

Label Shift has been widely believed to be harmful to the generalization performance of machine learning models. Researchers have proposed many approaches to mitigate the impact of the label shift, e.g., balancing the training data. However, these methods often consider the underparametrized regime, where the sample size is much larger than the data dimension. The research under the overparametrized regime is very limited. To bridge this gap, we propose a new asymptotic analysis of the Fisher Linear Discriminant classifier for binary classification with label shift. Specifically, we prove that there exists a phase transition phenomenon: Under certain overparametrized regime, the classifier trained using imbalanced data outperforms the counterpart with reduced balanced data. Moreover, we investigate the impact of regularization to the label shift: The aforementioned phase transition vanishes as the regularization becomes strong.

Context and Motivation Clustering is an unsupervised learning approach for the task of partitioning data into meaningful subsets. The huge literature on cluster analysis is difficult to survey in a few sentences, but a concise description of well-known approaches is offered by Hastie et al. (2009), Everitt et al. (2011), and King (2014). Examples of mainstream methods for clustering data include model-based (Bouveyron et al., 2019), similarity-based (MacQueen, 1967; Kaufman and Rousseeuw, 1987), and hierarchical clustering (Hastie et al., 2009, Section 14.3). In this paper we propose a novel model-based approach for cluster analysis that lies at the interface of model-based clustering (i.e., via mixture models) and similarity-based clustering (i.e., via K-means and K-medoids). The proposed approach aims to benefit from the flexibility and soundness of model-based clustering, while attempting to mitigate Pitfalls 1 and 2 below. Model-based clustering is a fast-evolving and intradisciplinary research topic as can be seen from the recent Handbook on Mixture Analysis (Fruhwirth-Schnatter et al., 2019) as well as the survey papers of Melnykov and Maitra (2010), McNicholas (2016), Gormley et al. (2023), and the references therein.

Multi-Task Learning (MTL) has shown its importance at user products for fast training, data efficiency, reduced overfitting etc. MTL achieves it by sharing the network parameters and training a network for multiple tasks simultaneously. However, MTL does not provide the solution, if each task needs training from a different dataset. In order to solve the stated problem, we have proposed an architecture named TreeDNN along with it's training methodology. TreeDNN helps in training the model with multiple datasets simultaneously, where each branch of the tree may need a different training dataset. We have shown in the results that TreeDNN provides competitive performance with the advantage of reduced ROM requirement for parameter storage and increased responsiveness of the system by loading only specific branch at inference time.

Arguments in argumentation systems cannot always be considered as standalone entities, requiring the consideration of the pieces of evidence they rely on. This evidence might have to be retrieved from external sources such as databases or the web, and each attempt to retrieve a piece of evidence comes with an associated cost. Moreover, a piece of evidence may be available in a given scenario but not in others, and this is not known beforehand. As a result, the collection of active arguments (whose entire set of evidence is available) that can be used by the argumentation machinery of the system may vary from one scenario to another. In this work, we consider an Abstract Argumentation Framework with Fallible Evidence that accounts for these issues, and propose a heuristic measure used as part of the acceptability calculus (specifically, for building pruned dialectical trees) with the aim of minimizing the evidence retrieval cost of the arguments involved in the reasoning process. We provide an algorithmic solution that is empirically tested against two baselines and formally show the correctness of our approach.

Runway's cofounders (from left) Anastasis Germanidis, Alejandro Matamala-Ortiz and Cristóbal Valenzuela are immigrants who met while studying as art students at New York University. Runway ML, one of the two startups behind the popular AI text-to-image model Stable Diffusion, has raised new funding at a $500 million valuation, Forbes has learned. Felicis is leading the new funding, the sources said, which comes on the heels of a boom in generative AI that has captured the public's attention in recent months thanks to releases that also include OpenAI's Dall-E and ChatGPT. Runway quickly emerged as one of the buzziest startups in the mix with its video editing software, for which the company has been releasing a bevy of generative AI features. For example, from a photo of a forest, a user can type a short text phrase into Runway's software and instantly conjure a lake or a castle among the trees.

The angular momentum of galaxies (galaxy spin) contains rich information about the initial condition of the Universe, yet it is challenging to efficiently measure the spin direction for the tremendous amount of galaxies that are being mapped by the ongoing and forthcoming cosmological surveys. We present a machine learning based classifier for the Z-wise vs S-wise spirals, which can help to break the degeneracy in the galaxy spin direction measurement. The proposed Chirality Equivariant Residual Network (CE-ResNet) is manifestly equivariant under a reflection of the input image, which guarantees that there is no inherent asymmetry between the Z-wise and S-wise probability estimators. We train the model with Sloan Digital Sky Survey (SDSS) images, with the training labels given by the Galaxy Zoo 1 (GZ1) project. A combination of data augmentation tricks are used during the training, making the model more robust to be applied to other surveys. We find a $\sim\!30\%$ increase of both types of spirals when Dark Energy Spectroscopic Instrument (DESI) images are used for classification, due to the better imaging quality of DESI. We verify that the $\sim\!7\sigma$ difference between the numbers of Z-wise and S-wise spirals is due to human bias, since the discrepancy drops to $<\!1.8\sigma$ with our CE-ResNet classification results. We discuss the potential systematics that are relevant to the future cosmological applications.

The majority of available text summarization datasets include short-form source documents that lack long-range causal and temporal dependencies, and often contain strong layout and stylistic biases. While relevant, such datasets will offer limited challenges for future generations of text summarization systems. We address these issues by introducing BookSum, a collection of datasets for long-form narrative summarization. Our dataset covers source documents from the literature domain, such as novels, plays and stories, and includes highly abstractive, human written summaries on three levels of granularity of increasing difficulty: paragraph-, chapter-, and book-level. The domain and structure of our dataset poses a unique set of challenges for summarization systems, which include: processing very long documents, non-trivial causal and temporal dependencies, and rich discourse structures. To facilitate future work, we trained and evaluated multiple extractive and abstractive summarization models as baselines for our dataset.