With the recent advance of representation learning algorithms on graphs (e.g., DeepWalk/GraphSage) and natural languages (e.g., Word2Vec/BERT) , the state-of-the art models can even achieve human-level performance over many downstream tasks, particularly for the task of node and sentence classification. However, most algorithms focus on large-scale models for static graphs and text corpus without considering the inherent dynamic characteristics or discovering the reasons behind the changes. This dissertation aims to efficiently model the dynamics in graphs (such as social networks and citation graphs) and understand the changes in texts (specifically news titles and personal biographies). To achieve this goal, we utilize the renowned Personalized PageRank algorithm to create effective dynamic network embeddings for evolving graphs. Our proposed approaches significantly improve the running time and accuracy for both detecting network abnormal intruders and discovering entity meaning shifts over large-scale dynamic graphs. For text changes, we analyze the post-publication changes in news titles to understand the intents behind the edits and discuss the potential impact of titles changes from information integrity perspective. Moreover, we investigate self-presented occupational identities in Twitter users' biographies over five years, investigating job prestige and demographics effects in how people disclose jobs, quantifying over-represented jobs and their transitions over time.

Knowledge representation and reasoning systems represent knowledge as collections of facts and rules. KRRs can represent complex concepts and relations, and they can query and manipulate information in sophisticated ways. Unfortunately, the KRR technology has been hindered by the fact that specifying the requisite knowledge requires skills that most domain experts do not have, and professional knowledge engineers are hard to find. Some recent CNL-based approaches, such as the Knowledge Authoring Logic Machine (KALM), have shown to have very high accuracy compared to others, and a natural question is to what extent the CNL restrictions can be lifted. Besides the CNL restrictions, KALM has limitations in terms of the types of knowledge it can represent. To address these issues, we propose an extension of KALM called KALM for Factual Language (KALMF). KALMF uses a neural parser for natural language, MS, to parse what we call factual English sentences, which require little grammar training to use. Building upon KALMF, we propose KALM for Rules and Actions (KALMR), to represent and reason with rules and actions. Furthermore, we identify the reasons behind the slow speed of KALM and make optimizations to address this issue. Our evaluation using multiple benchmarks shows that our approaches achieve a high level of correctness on fact and query authoring (95%) and on rule authoring (100%). When used for authoring and reasoning with actions, our approach achieves more than 99.3% correctness, demonstrating its effectiveness in enabling more sophisticated knowledge representation and reasoning. We also illustrate the logical reasoning capabilities of our approach by drawing attention to the problems faced by the famous AI, ChatGPT. Finally, the evaluation of the newly proposed speed optimization points not only to a 68% runtime improvement but also yields better accuracy of the overall system.

The mental health of social media users has started more and more to be put at risk by harmful, hateful, and offensive content. In this paper, we propose \textsc{StopHC}, a harmful content detection and mitigation architecture for social media platforms. Our aim with \textsc{StopHC} is to create more secure online environments. Our solution contains two modules, one that employs deep neural network architecture for harmful content detection, and one that uses a network immunization algorithm to block toxic nodes and stop the spread of harmful content. The efficacy of our solution is demonstrated by experiments conducted on two real-world datasets.

The recent introduction of large language models (LLMs) has revolutionized the field of robotics by enabling contextual reasoning and intuitive human-robot interaction in domains as varied as manipulation, locomotion, and self-driving vehicles. When viewed as a stand-alone technology, LLMs are known to be vulnerable to jailbreaking attacks, wherein malicious prompters elicit harmful text by bypassing LLM safety guardrails. To assess the risks of deploying LLMs in robotics, in this paper, we introduce RoboPAIR, the first algorithm designed to jailbreak LLM-controlled robots. Unlike existing, textual attacks on LLM chatbots, RoboPAIR elicits harmful physical actions from LLM-controlled robots, a phenomenon we experimentally demonstrate in three scenarios: (i) a white-box setting, wherein the attacker has full access to the NVIDIA Dolphins self-driving LLM, (ii) a gray-box setting, wherein the attacker has partial access to a Clearpath Robotics Jackal UGV robot equipped with a GPT-4o planner, and (iii) a black-box setting, wherein the attacker has only query access to the GPT-3.5-integrated Unitree Robotics Go2 robot dog. In each scenario and across three new datasets of harmful robotic actions, we demonstrate that RoboPAIR, as well as several static baselines, finds jailbreaks quickly and effectively, often achieving 100% attack success rates. Our results reveal, for the first time, that the risks of jailbroken LLMs extend far beyond text generation, given the distinct possibility that jailbroken robots could cause physical damage in the real world. Indeed, our results on the Unitree Go2 represent the first successful jailbreak of a deployed commercial robotic system. Addressing this emerging vulnerability is critical for ensuring the safe deployment of LLMs in robotics. Additional media is available at: https://robopair.org

TikTok user and travel agent Taylor Moore shared details of her plane ride next to Taylor Swift's dad, including his proud papa moments and approval of Travis Kelce. The Beatles, Taylor Swift and Beyoncé are facing off at the 2025 Grammy Awards. On Friday, the Recording Academy released its full list of Grammy nominations, and the Beatles earned their first nod since 1997 for their latest song, "Now and Then." The Fab Four also earned a nomination for the same song in the best rock performance category. The Beatles' last new song, the AI-assisted "Now and Then," was released in 2023.

Apple's even tinier Mac mini is here -- with M4 power. It won't shock you to hear the M4 Pro is very fast, but the Mac mini comes with 16 gigs of RAM as standard too. The base Mac mini has an M4 chip sports a 10-core CPU (four high-performance cores and six high efficiency), a 10-core GPU and a 16-core Neural Engine. For 1,399, you can bump up to the dramatically more powerful M4 Pro chip (like our review unit), featuring a 14-core CPU (10 high-performance and four high efficiency) and 20-core GPU. Its Geekbench 6 and Cinebench scores still beat most of the computers we've tested this year, and its GPU is fast enough for solid 1080p 60 fps gameplay.

Recent advancements in Large Language Models (LLMs) have significantly enhanced their capacity to process long contexts. However, effectively utilizing this long context remains a challenge due to the issue of distraction, where irrelevant information dominates lengthy contexts, causing LLMs to lose focus on the most relevant segments. To address this, we propose a novel training method that enhances LLMs' ability to discern relevant information through a unique combination of retrieval-based data augmentation and contrastive learning. Specifically, during fine-tuning with long contexts, we employ a retriever to extract the most relevant segments, serving as augmented inputs. We then introduce an auxiliary contrastive learning objective to explicitly ensure that outputs from the original context and the retrieved sub-context are closely aligned. Extensive experiments on long single-document and multi-document QA benchmarks demonstrate the effectiveness of our proposed method.

Multimodal foundation models, such as Gemini and ChatGPT, have revolutionized human-machine interactions by seamlessly integrating various forms of data. Developing a universal spoken language model that comprehends a wide range of natural language instructions is critical for bridging communication gaps and facilitating more intuitive interactions. However, the absence of a comprehensive evaluation benchmark poses a significant challenge. We present Dynamic-SUPERB Phase-2, an open and evolving benchmark for the comprehensive evaluation of instruction-based universal speech models. Building upon the first generation, this second version incorporates 125 new tasks contributed collaboratively by the global research community, expanding the benchmark to a total of 180 tasks, making it the largest benchmark for speech and audio evaluation. While the first generation of Dynamic-SUPERB was limited to classification tasks, Dynamic-SUPERB Phase-2 broadens its evaluation capabilities by introducing a wide array of novel and diverse tasks, including regression and sequence generation, across speech, music, and environmental audio. Evaluation results indicate that none of the models performed well universally. SALMONN-13B excelled in English ASR, while WavLLM demonstrated high accuracy in emotion recognition, but current models still require further innovations to handle a broader range of tasks. We will soon open-source all task data and the evaluation pipeline.

Separating disinformation from fact on the web has long challenged both the search and the reasoning powers of humans. We show that the reasoning power of large language models (LLMs) and the retrieval power of modern search engines can be combined to automate this process and explainably verify claims. We integrate LLMs and search under a multi-hop evidence pursuit strategy. This strategy generates an initial question based on an input claim using a sequence to sequence model, searches and formulates an answer to the question, and iteratively generates follow-up questions to pursue the evidence that is missing using an LLM. We demonstrate our system on the FEVER 2024 (AVeriTeC) shared task. Compared to a strategy of generating all the questions at once, our method obtains .045 higher label accuracy and .155 higher AVeriTeC score (evaluating the adequacy of the evidence). Through ablations, we show the importance of various design choices, such as the question generation method, medium-sized context, reasoning with one document at a time, adding metadata, paraphrasing, reducing the problem to two classes, and reconsidering the final verdict. Our submitted system achieves .510 AVeriTeC score on the dev set and .477 AVeriTeC score on the test set.

Large Language Models (LLMs), with gradually improving reading comprehension and reasoning capabilities, are being applied to a range of complex language tasks, including the automatic generation of language data for various purposes. However, research on applying LLMs for automatic data generation in low-resource languages like Vietnamese is still underdeveloped and lacks comprehensive evaluation. In this paper, we explore the use of LLMs for automatic data generation for the Vietnamese fact-checking task, which faces significant data limitations. Specifically, we focus on fact-checking data where claims are synthesized from multiple evidence sentences to assess the information synthesis capabilities of LLMs. We develop an automatic data construction process using simple prompt techniques on LLMs and explore several methods to improve the quality of the generated data. To evaluate the quality of the data generated by LLMs, we conduct both manual quality assessments and performance evaluations using language models. Experimental results and manual evaluations illustrate that while the quality of the generated data has significantly improved through fine-tuning techniques, LLMs still cannot match the data quality produced by humans.