In recent years, the Natural Language Inference (NLI) task has garnered significant attention, with new datasets and models achieving near human-level performance on it. However, the full promise of NLI -- particularly that it learns knowledge that should be generalizable to other downstream NLP tasks -- has not been realized. In this paper, we study this unfulfilled promise from the lens of two downstream tasks: question answering (QA), and text summarization. We conjecture that a key difference between the NLI datasets and these downstream tasks concerns the length of the premise; and that creating new long premise NLI datasets out of existing QA datasets is a promising avenue for training a truly generalizable NLI model. We validate our conjecture by showing competitive results on the task of QA and obtaining the best reported results on the task of Checking Factual Correctness of Summaries.

AI Daily Roundup starts today! We are covering the top updates from around the world. The updates will feature state-of-the-art capabilities in artificial intelligence, Machine Learning, Robotic Process Automation, Fintech and human-system interactions. We will cover the role of AI Daily Roundup and their application in various industries and daily lives. SEMrush, leading online visibility management, and content marketing SaaS platform, has further expanded its ecosystem with the acquisition of a 100% stake in fast-growing public relations SaaS startup Prowly.com.

Last week, it was Xbox's turn, and now we have a price for the new PlayStation. Sony has set a $400/$500 split for the Digital Edition vs. standard PS5, charging $100 extra for the privilege of a disc drive for physical media. If you need a reason to splash for the disc version, look no further than the new $70 benchmark for many new titles -- disc games might be a little easier to find used or on sale. Retailers like Amazon, Best Buy, Walmart and GameStop started taking pre-orders shortly after Sony's event ended, but if you missed out on the first wave, keep checking back -- a few people have reported success hours after they supposedly sold out. Assuming you can secure a day-one purchase, you can expect it to see it November 12th in the US, Canada, Japan, Mexico, Australia, New Zealand and South Korea, before the PS5 launches everywhere else on November 19th.

In Part 1 of this article on AI and web development, the impacts of AI on web development is mentioned. One of these significant impacts is AI-powered web builders that elevates website design processes dramatically. Through also known as Artificial Design Intelligence ADI systems, the usually complicated, time-consuming, prone to human error process of User Interface (UI) design process is streamlined. It has changed the way web developers create and maintain websites and applications, both for desktop and mobile. Here are 7 of the best and leading web builders powered by AI shaking the web development industry today.

Earlier this year, the independent research organisation of which I am the Director, London-based Ada Lovelace Institute, hosted a panel at the world's largest AI conference, CogX, called The Ethics Panel to End All Ethics Panels. The title referenced both a tongue-in-cheek effort at self-promotion, and a very real need to put to bed the seemingly endless offering of panels, think-pieces, and government reports preoccupied with ruminating on the abstract ethical questions posed by AI and new data-driven technologies. We had grown impatient with conceptual debates and high-level principles. And we were not alone. It supersedes the two waves that came before it: the first wave, defined by principles and dominated by philosophers, and the second wave, led by computer scientists and geared towards technical fixes.

Structured belief states are crucial for user goal tracking and database query in task-oriented dialog systems. However, training belief trackers often requires expensive turn-level annotations of every user utterance. In this paper we aim at alleviating the reliance on belief state labels in building end-to-end dialog systems, by leveraging unlabeled dialog data towards semi-supervised learning. We propose a probabilistic dialog model, called the LAtent BElief State (LABES) model, where belief states are represented as discrete latent variables and jointly modeled with system responses given user inputs. Such latent variable modeling enables us to develop semi-supervised learning under the principled variational learning framework. Furthermore, we introduce LABES-S2S, which is a copy-augmented Seq2Seq model instantiation of LABES. In supervised experiments, LABES-S2S obtains strong results on three benchmark datasets of different scales. In utilizing unlabeled dialog data, semi-supervised LABES-S2S significantly outperforms both supervised-only and semi-supervised baselines. Remarkably, we can reduce the annotation demands to 50% without performance loss on MultiWOZ.

The aim of all Question Answering (QA) systems is to be able to generalize to unseen questions. Current supervised methods are reliant on expensive data annotation. Moreover, such annotations can introduce unintended annotator bias which makes systems focus more on the bias than the actual task. In this work, we propose Knowledge Triplet Learning (KTL), a self-supervised task over knowledge graphs. We propose heuristics to create synthetic graphs for commonsense and scientific knowledge. We propose methods of how to use KTL to perform zero-shot QA and our experiments show considerable improvements over large pre-trained transformer models.

In complex areas such as law and science, knowledge has been in centuries formulated by primarily describing prototypical instances and properties, and then by overriding the general theory to include possible exceptions. For example, many laws are formulated by adding new norms that, in case of conflicts, may partially or completely override the previous ones. Similarly, biologists have been incrementally introducing exceptions to general properties. For instance, the human heart is usually located in the left-hand half of the thorax. Still there are exceptional individuals, with so-called situs inversus, whose heart is located on the opposite side. Eukariotic cells are those with a proper nucleus, by definition. Still they comprise mammalian red blood cells, that in their mature stage have no nucleus.

Inducing a meaningful structural representation from one or a set of dialogues is a crucial but challenging task in computational linguistics. Advancement made in this area is critical for dialogue system design and discourse analysis. It can also be extended to solve grammatical inference. In this work, we propose to incorporate structured attention layers into a Variational Recurrent Neural Network (VRNN) model with discrete latent states to learn dialogue structure in an unsupervised fashion. Compared to a vanilla VRNN, structured attention enables a model to focus on different parts of the source sentence embeddings while enforcing a structural inductive bias. Experiments show that on two-party dialogue datasets, VRNN with structured attention learns semantic structures that are similar to templates used to generate this dialogue corpus. While on multi-party dialogue datasets, our model learns an interactive structure demonstrating its capability of distinguishing speakers or addresses, automatically disentangling dialogues without explicit human annotation.

Multi-label classification (MLC) studies the problem where each instance is associated with multiple relevant labels, which leads to the exponential growth of output space. MLC encourages a popular framework named label compression (LC) for capturing label dependency with dimension reduction. Nevertheless, most existing LC methods failed to consider the influence of the feature space or misguided by original problematic features, so that may result in performance degeneration. In this paper, we present a compact learning (CL) framework to embed the features and labels simultaneously and with mutual guidance. The proposal is a versatile concept, hence the embedding way is arbitrary and independent of the subsequent learning process. Following its spirit, a simple yet effective implementation called compact multi-label learning (CMLL) is proposed to learn a compact low-dimensional representation for both spaces. CMLL maximizes the dependence between the embedded spaces of the labels and features, and minimizes the loss of label space recovery concurrently. Theoretically, we provide a general analysis for different embedding methods. Practically, we conduct extensive experiments to validate the effectiveness of the proposed method.