One of the most important features of real-world networks is their community structure, as it reveals the internal organization of nodes [10]. In social networks, communities may represent groups by interest; in citation networks, they correspond to related papers; on the Web, communities are formed by pages on related topics, etc. Being able to identify communities in a network could help us to exploit this network more effectively. Grouping like-minded users or similar-looking items together is important for a wide range of applications, including controlling epidemics [12], recommendation systems, anomaly or outlier detection, fraud detection, rumor or fake news detection, etc. [16]. There is also growing literature introducing community-aware centrality measures that exploit both local and global properties of networks [8, 35]. For more discussion around various aspects of mining complex networks, see for example, [31, 23].

Controlling the text generated by language models and customizing the content has been a long-standing challenge. Existing prompting techniques proposed in pursuit of providing control are task-specific and lack generality; this provides overwhelming choices for non-expert users to find a suitable method for their task. The effort associated with those techniques, such as in writing examples, explanations, instructions, etc. further limits their adoption among non-expert users. In this paper, we propose a simple prompting strategy HELP ME THINK where we encourage GPT3 to help non-expert users by asking a set of relevant questions and leveraging user answers to execute the task. We demonstrate the efficacy of our technique HELP ME THINK on a variety of tasks. Specifically, we focus on tasks that are hard for average humans and require significant thinking to perform. We hope our work will encourage the development of unconventional ways to harness the power of large language models.

Language models demonstrate both quantitative improvement and new qualitative capabilities with increasing scale. Despite their potentially transformative impact, these new capabilities are as yet poorly characterized. In order to inform future research, prepare for disruptive new model capabilities, and ameliorate socially harmful effects, it is vital that we understand the present and near-future capabilities and limitations of language models. To address this challenge, we introduce the Beyond the Imitation Game benchmark (BIG-bench). BIG-bench currently consists of 204 tasks, contributed by 450 authors across 132 institutions. Task topics are diverse, drawing problems from linguistics, childhood development, math, common-sense reasoning, biology, physics, social bias, software development, and beyond. BIG-bench focuses on tasks that are believed to be beyond the capabilities of current language models. We evaluate the behavior of OpenAI's GPT models, Google-internal dense transformer architectures, and Switch-style sparse transformers on BIG-bench, across model sizes spanning millions to hundreds of billions of parameters. In addition, a team of human expert raters performed all tasks in order to provide a strong baseline. Findings include: model performance and calibration both improve with scale, but are poor in absolute terms (and when compared with rater performance); performance is remarkably similar across model classes, though with benefits from sparsity; tasks that improve gradually and predictably commonly involve a large knowledge or memorization component, whereas tasks that exhibit "breakthrough" behavior at a critical scale often involve multiple steps or components, or brittle metrics; social bias typically increases with scale in settings with ambiguous context, but this can be improved with prompting.

Harvey Castro talks about how AI cold be used in cold cases and the symbiotic relationship between AI and a detective. There's no question that artificial intelligence is changing our lives. A bot that sounds almost human can author your emails, teach you a new language, book your trips or even be your friend. Check out direct links to try those out here. One woman I spoke with on my national radio show even married her AI companion.

Social media platforms are essential outlets for expressing opinions, providing a valuable resource for capturing public viewpoints via text analytics. However, for many users, passive browsing is their preferred mode of interaction, leading to their perspectives being overlooked by text analytics methods. Meanwhile, social media polls have emerged as a practical feature for gathering public opinions, allowing post authors to pose questions with pre-defined answer options for readers to vote on. To broaden the benefits of polls for posts without them, this article explores the automatic generation of a poll from a social media post by leveraging cutting-edge natural language generation (NLG) techniques. However, existing NLG techniques, primarily developed for general-domain texts, may be ineffective when applied to noisy social media data, which often feature implicit context-question-answer relations. To tackle these challenges, we enrich a post context with its comments and propose a novel unified poll generation framework called UniPoll. It employs prompt tuning with multi-objective optimization to bolster the connection exploration between contexts (posts and comments) and polls (questions and answers). Experimental comparisons on a large-scale Chinese Weibo dataset show that UniPoll significantly outperforms T5, the state-of-the-art NLG model, which generates question and answer separately. Comprehensive qualitative and quantitative analyses further underscore the superiority of UniPoll through various evaluation lenses.

Recently, denoising diffusion models have demonstrated remarkable performance among generative models in various domains. However, in the speech domain, the application of diffusion models for synthesizing time-varying audio faces limitations in terms of complexity and controllability, as speech synthesis requires very high-dimensional samples with long-term acoustic features. To alleviate the challenges posed by model complexity in singing voice synthesis, we propose HiddenSinger, a high-quality singing voice synthesis system using a neural audio codec and latent diffusion models. To ensure high-fidelity audio, we introduce an audio autoencoder that can encode audio into an audio codec as a compressed representation and reconstruct the high-fidelity audio from the low-dimensional compressed latent vector. Subsequently, we use the latent diffusion models to sample a latent representation from a musical score. In addition, our proposed model is extended to an unsupervised singing voice learning framework, HiddenSinger-U, to train the model using an unlabeled singing voice dataset. Experimental results demonstrate that our model outperforms previous models in terms of audio quality. Furthermore, the HiddenSinger-U can synthesize high-quality singing voices of speakers trained solely on unlabeled data.

The invention of modern displays has enhanced the viewer experience for any kind of content: ranging from sports to movies in 8K high-definition resolution. However, older content developed for CRT or early Plasma screen TVs has become outdated quickly and no longer meets current aspect ratio and resolution standards. In this paper, we explore whether we can solve this problem with the use of diffusion models to adapt old content to meet contemporary expectations. We explore the ability to combine multiple independent computer vision tasks to attempt to solve the problem of expanding aspect ratios of old animated content such that the new content would be indistinguishable from the source material to a brand-new viewer. These existing capabilities include Stable Diffusion, Content-Aware Scene Detection, Object Detection, and Key Point Matching. We were able to successfully chain these tasks together in a way that generated reasonable outputs, however, future work needs to be done to improve and expand the application to non-animated content as well.

This paper explores how AI-owners can develop safeguards for AI-generated content by drawing from established codes of conduct and ethical standards in other content-creation industries. It delves into the current state of ethical awareness on Large Language Models (LLMs). By dissecting the mechanism of content generation by LLMs, four key areas (upstream/downstream and at user prompt/answer), where safeguards could be effectively applied, are identified. A comparative analysis of these four areas follows and includes an evaluation of the existing ethical safeguards in terms of cost, effectiveness, and alignment with established industry practices. The paper's key argument is that existing IT-related ethical codes, while adequate for traditional IT engineering, are inadequate for the challenges posed by LLM-based content generation. Drawing from established practices within journalism, we propose potential standards for businesses involved in distributing and selling LLM-generated content. Finally, potential conflicts of interest between dataset curation at upstream and ethical benchmarking downstream are highlighted to underscore the need for a broader evaluation beyond mere output. This study prompts a nuanced conversation around ethical implications in this rapidly evolving field of content generation.

With the growing importance of detecting misinformation, many studies have focused on verifying factual claims by retrieving evidence. However, canonical fact verification tasks do not apply to catching subtle differences in factually consistent claims, which might still bias the readers, especially on contentious political or economic issues. Our underlying assumption is that among the trusted sources, one's argument is not necessarily more true than the other, requiring comparison rather than verification. In this study, we propose ClaimDiff, a novel dataset that primarily focuses on comparing the nuance between claim pairs. In ClaimDiff, we provide 2,941 annotated claim pairs from 268 news articles. We observe that while humans are capable of detecting the nuances between claims, strong baselines struggle to detect them, showing over a 19% absolute gap with the humans. We hope this initial study could help readers to gain an unbiased grasp of contentious issues through machine-aided comparison.

The monocular visual-inertial odometry (VIO) based on the direct method can leverage all available pixels in the image to simultaneously estimate the camera motion and reconstruct the denser map of the scene in real time. However, the direct method is sensitive to photometric changes, which can be compensated by introducing geometric information in the environment. In this paper, we propose a monocular direct sparse visual-inertial odometry, which exploits the planar regularities (PVI-DSO). Our system detects the planar regularities from the 3D mesh built on the estimated map points. To improve the pose estimation accuracy with the geometric information, a tightly coupled coplanar constraint expression is used to express photometric error in the direct method. Additionally, to improve the optimization efficiency, we elaborately derive the analytical Jacobian of the linearization form for the coplanar constraint. Finally, the inertial measurement error, coplanar point photometric error, non-coplanar photometric error, and prior error are added into the optimizer, which simultaneously improves the pose estimation accuracy and mesh itself. We verified the performance of the whole system on simulation and real-world datasets. Extensive experiments have demonstrated that our system outperforms the state-of-the-art counterparts.