Automatically identifying rumors from online social media especially microblogging websites is an important research issue. Most of existing work for rumor detection focuses on modeling features related to microblog contents, users and propagation patterns, but ignore the importance variation these context during message over time. In this study, we propose a novel approach capture temporal characteristics based time series rumor's lifecycle, which technique applied incorporate various...

Automatic rumor detection is technically very challenging. In this work, we try to learn discriminative features from tweets content by following their non-sequential propagation structure and generate more powerful representations for identifying different type of rumors. We propose two recursive neural models based on a bottom-up top-down tree-structured networks representation learning classification, which naturally conform the layout tweets. Results public Twitter datasets demonstrate...

How fake news goes viral via social media? does its propagation pattern differ from real stories? In this paper, we attempt to address the problem of identifying rumors, i.e., information, out microblog posts based on their structure. We firstly model diffusion with trees, which provide valuable clues how an original message is transmitted and developed over time. then propose a kernel-based method called Propagation Tree Kernel, captures high-order patterns differentiating different types...

Rumors can cause devastating consequences to individual and/or society. Analysis shows that widespread of rumors typically results from deliberately promoted information campaigns which aim shape collective opinions on the concerned news events. In this paper, we attempt fight such chaos with itself make automatic rumor detection more robust and effective. Our idea is inspired by adversarial learning method originated Generative Adversarial Networks (GAN). We propose a GAN-style approach,...

In recent years, an unhealthy phenomenon characterized as the massive spread of fake news or unverified information (i.e., rumors) has become increasingly a daunting issue in human society. The rumors commonly originate from social media outlets, primarily microblogging platforms, being viral afterwards by wild, willful propagation via large number participants. It is observed that rumorous posts often trigger versatile, mostly controversial stances among participating users. Thus,...

Claim verification is generally a task of verifying the veracity given claim, which critical to many downstream applications. It cumbersome and inefficient for human fact-checkers find consistent pieces evidence, from solid verdict could be inferred against claim. In this paper, we propose novel end-to-end hierarchical attention network focusing on learning represent coherent evidence as well their semantic relatedness with Our model consists three main components: 1) A coherence-based layer...

Federated learning has emerged as an important paradigm for training machine models in different domains. For graph-level tasks such graph classification, graphs can also be regarded a special type of data samples, which collected and stored separate local systems. Similar to other domains, multiple systems, each holding small set graphs, may benefit from collaboratively powerful mining model, the popular neural networks (GNNs). To provide more motivation towards endeavors, we analyze...

Rumor spread in social media severely jeopardizes the credibility of online content. Thus, automatic debunking rumors is great importance to keep a healthy environment. While facing dubious claim, people often dispute its truthfulness sporadically their posts containing various cues, which can form useful evidence with long-distance dependencies. In this work, we propose learn discriminative features from microblog by following non-sequential propagation structure and generate more powerful...

In the era of information explosion, named entity recognition (NER) has attracted widespread attention in field natural language processing, as it is fundamental to extraction. Recently, methods NER based on representation learning, e.g., character embedding and word embedding, have demonstrated promising results. However, existing models only consider partial features derived from words or characters while failing integrate semantic syntactic information, capitalization, inter-word...

The diffusion of rumors on social media generally follows a propagation tree structure, which provides valuable clues how an original message is transmitted and responded by users over time. Recent studies reveal that rumor verification stance detection are two relevant tasks can jointly enhance each other despite their differences. For example, be debunked cross-checking the stances conveyed posts, also conditioned nature rumor. However, typically requires large training set labeled at post...

Electronic health records (EHRs) are a valuable source of information that can aid in understanding patient's condition and making informed healthcare decisions. However, modelling longitudinal EHRs with heterogeneous is challenging task. Although recurrent neural networks (RNNs), which current artificial intelligence (AI) models, have the capability to capture information, their explanatory power limited. Predictive clustering recent development this field, provides cluster-level...

In this work, we use an extensive empirical database of errors induced by write, read, and erase operations to develop a comprehensive understanding the error behavior flash memories. Error characterization MLC SLC is given on block, page, bit level. Based our in flash, propose error-correcting scheme which outperforms conventional BCH code. We compare several schemes block as block. Finally, implementation two-write WOM-codes well BER for first second write.

Ubiquitous use of social media such as microblogging platforms brings about ample opportunities for the false information to diffuse online. It is very important not just determine veracity but also authenticity users who spread information, especially in time-critical situations like real-world emergencies, where urgent measures have be taken stopping fake information. In this work, we propose a novel machine learning based approach automatic identification spreading rumorous by leveraging...

Tensor factorization has been demonstrated as an efficient approach for computational phenotyping, where massive electronic health records (EHRs) are converted to concise and meaningful clinical concepts. While distributing the tensor tasks local sites can avoid direct data sharing, it still requires exchange of intermediary results which could reveal sensitive patient information. Therefore, challenge is how jointly decompose under rigorous principled privacy constraints, while support...

Rumors are manufactured with no respect for accuracy, but can circulate quickly and widely by “word-of-post” through social media conversations. Conversation tree encodes important information indicative of the credibility rumor. Existing conversation-based techniques rumor detection either just strictly follow edges or treat all posts fully-connected during feature learning. In this paper, we propose a novel model based on transformer to better utilize user interactions in dialogue where...

Rumors are rampant in the era of social media. Conversation structures provide valuable clues to differentiate between real and fake claims. However, existing rumor detection methods either limited strict relation user responses or oversimplify conversation structure. In this study, substantially reinforces interaction opinions while alleviating negative impact imposed by irrelevant posts, we first represent thread as an undirected graph. We then present a Claim-guided Hierarchical Graph...

The COVID-19 pandemic poses a great threat to global public health. Meanwhile, there is massive misinformation associated with the which advocates unfounded or unscientific claims. Even major social media and news outlets have made an extra effort in debunking misinformation, most of fact-checking information English, whereas some unmoderated still circulating other languages, threatening health less-informed people immigrant communities developing countries. In this paper, we make first...

Electronic health records (EHRs) contain diverse patient information, including medical notes, clinical events, and laboratory test results. Integrating this multimodal data can improve disease diagnoses using deep learning models. However, effectively combining different modalities for diagnosis remains challenging. Previous approaches, such as attention mechanisms contrastive learning, have attempted to address but do not fully integrate the into a unified feature space. This paper...

Image Captioning is a traditional vision-and-language task that aims to generate the language description of an image. Recent studies focus on scaling up model size and number training data, which significantly increase cost training. Different these heavy-cost models, we introduce lightweight image captioning framework (I-Tuning), contains small trainable parameters. We design novel I-Tuning cross-attention module connect non-trainable pre-trained decoder GPT2 vision encoder CLIP-ViT. Since...

The proliferation of misinformation, such as rumors on social media, has drawn significant attention, prompting various expressions stance among users. Although rumor detection and are distinct tasks, they can complement each other. Rumors be identified by cross-referencing stances in related posts, influenced the nature rumor. However, existing methods often require post-level annotations, which costly to obtain. We propose a novel LLM-enhanced MIL approach jointly predict post claim class...

The proliferation of misinformation, such as rumors on social media, has drawn significant attention, prompting various expressions stance among users. Although rumor detection and are distinct tasks, they can complement each other. Rumors be identified by cross-referencing stances in related posts, influenced the nature rumor. However, existing methods often require post-level annotations, which costly to obtain. We propose a novel LLM-enhanced MIL approach jointly predict post claim class...

The ability of large language models (LLMs) to follow instructions is crucial for their practical applications, yet the underlying mechanisms remain poorly understood. This paper presents a novel framework that leverages sparse autoencoders (SAE) interpret how instruction following works in these models. We demonstrate features we identify can effectively steer model outputs align with given instructions. Through analysis SAE latent activations, specific latents responsible behavior. Our...