Anomaly detection plays a crucial role in various domains, from cybersecurity to industrial systems. However, traditional centralized approaches often encounter challenges related to data privacy. In this context, Federated Learning emerges as a promising solution. This work introduces two innovative algorithms--Ensemble SVDD and Support Vector Election--that leverage Support Vector Machines for anomaly detection in a federated setting. In comparison with the Neural Networks typically used in within Federated Learning, these new algorithms emerge as potential alternatives, as they can operate effectively with small datasets and incur lower computational costs. The novel algorithms are tested in various distributed system configurations, yielding promising initial results that pave the way for further investigation.

This paper presents Diffusion Forcing, a new training paradigm where a diffusion model is trained to denoise a set of tokens with independent per-token noise levels. We apply Diffusion Forcing to sequence generative modeling by training a causal next-token prediction model to generate one or several future tokens without fully diffusing past ones. Our approach is shown to combine the strengths of next-token prediction models, such as variable-length generation, with the strengths of full-sequence diffusion models, such as the ability to guide sampling to desirable trajectories. Our method offers a range of additional capabilities, such as (1) rolling-out sequences of continuous tokens, such as video, with lengths past the training horizon, where baselines diverge and (2) new sampling and guiding schemes that uniquely profit from Diffusion Forcing's variable-horizon and causal architecture, and which lead to marked performance gains in decision-making and planning tasks. In addition to its empirical success, our method is proven to optimize a variational lower bound on the likelihoods of all subsequences of tokens drawn from the true joint distribution. Project website: https://boyuan.space/diffusion-forcing

Recent efforts to address hallucinations in Large Language Models (LLMs) have focused on attributed text generation, which supplements generated texts with citations of supporting sources for post-generation fact-checking and corrections. Yet, these citations often point to entire documents or paragraphs, burdening users with extensive verification work. In this paper, we introduce a locally-attributable text generation approach, prioritizing concise attributions. Our method, named "Attribute First, then Generate", breaks down the conventional end-to-end generation process into three intuitive steps: content selection, sentence planning, and sequential sentence generation. By initially identifying relevant source segments ("select first") and then conditioning the generation process on them ("then generate"), we ensure these segments also act as the output's fine-grained attributions ("select" becomes "attribute"). Tested on Multi-document Summarization and Long-form Question-answering, our method not only yields more concise citations than the baselines but also maintains - and in some cases enhances - both generation quality and attribution accuracy. Furthermore, it significantly reduces the time required for fact verification by human assessors.

Generative AI raises many societal concerns such as boosting disinformation and propaganda campaigns. Watermarking AI-generated content is a key technology to address these concerns and has been widely deployed in industry. However, watermarking is vulnerable to removal attacks and forgery attacks. In this work, we propose the first image watermarks with certified robustness guarantees against removal and forgery attacks. Our method leverages randomized smoothing, a popular technique to build certifiably robust classifiers and regression models. Our major technical contributions include extending randomized smoothing to watermarking by considering its unique characteristics, deriving the certified robustness guarantees, and designing algorithms to estimate them. Moreover, we extensively evaluate our image watermarks in terms of both certified and empirical robustness. Our code is available at \url{https://github.com/zhengyuan-jiang/Watermark-Library}.

We propose using performance metrics derived from zero-failure testing to assess binary classifiers. The principal characteristic of the proposed approach is the asymmetric treatment of the two types of error. In particular, we construct a test set consisting of positive and negative samples, set the operating point of the binary classifier at the lowest value that will result to correct classifications of all positive samples, and use the algorithm's success rate on the negative samples as a performance measure. A property of the proposed approach, setting it apart from other commonly used testing methods, is that it allows the construction of a series of tests of increasing difficulty, corresponding to a nested sequence of positive sample test sets. We illustrate the proposed method on the problem of age estimation for determining whether a subject is above a legal age threshold, a problem that exemplifies the asymmetry of the two types of error. Indeed, misclassifying an under-aged subject is a legal and regulatory issue, while misclassifications of people above the legal age is an efficiency issue primarily concerning the commercial user of the age estimation system.

Large Language Models (LLMs) are powerful zero-shot assessors used in real-world situations such as assessing written exams and benchmarking systems. Despite these critical applications, no existing work has analyzed the vulnerability of judge-LLMs to adversarial manipulation. This work presents the first study on the adversarial robustness of assessment LLMs, where we demonstrate that short universal adversarial phrases can be concatenated to deceive judge LLMs to predict inflated scores. Since adversaries may not know or have access to the judge-LLMs, we propose a simple surrogate attack where a surrogate model is first attacked, and the learned attack phrase then transferred to unknown judge-LLMs. We propose a practical algorithm to determine the short universal attack phrases and demonstrate that when transferred to unseen models, scores can be drastically inflated such that irrespective of the assessed text, maximum scores are predicted. It is found that judge-LLMs are significantly more susceptible to these adversarial attacks when used for absolute scoring, as opposed to comparative assessment. Our findings raise concerns on the reliability of LLM-as-a-judge methods, and emphasize the importance of addressing vulnerabilities in LLM assessment methods before deployment in high-stakes real-world scenarios.

The open-ended nature of language generation makes the evaluation of autoregressive large language models (LLMs) challenging. One common evaluation approach uses multiple-choice questions (MCQ) to limit the response space. The model is then evaluated by ranking the candidate answers by the log probability of the first token prediction. However, first-tokens may not consistently reflect the final response output, due to model's diverse response styles such as starting with "Sure" or refusing to answer. Consequently, MCQ evaluation is not indicative of model behaviour when interacting with users. But by how much? We evaluate how aligned first-token evaluation is with the text output along several dimensions, namely final option choice, refusal rate, choice distribution and robustness under prompt perturbation. Our results show that the two approaches are severely misaligned on all dimensions, reaching mismatch rates over 60%. Models heavily fine-tuned on conversational or safety data are especially impacted. Crucially, models remain misaligned even when we increasingly constrain prompts, i.e., force them to start with an option letter or example template. Our findings i) underscore the importance of inspecting the text output as well and ii) caution against relying solely on first-token evaluation.

Planetary exploration requires traversal in environments with rugged terrains. In addition, Mars rovers and other planetary exploration robots often carry sensitive scientific experiments and components onboard, which must be protected from mechanical harm. This paper deals with an active suspension system focused on chassis stabilisation and an efficient traversal method while encountering unavoidable obstacles. Soft Actor-Critic (SAC) was applied along with Proportional Integral Derivative (PID) control to stabilise the chassis and traverse large obstacles at low speeds. The model uses the rover's distance from surrounding obstacles, the height of the obstacle, and the chassis' orientation to actuate the control links of the suspension accurately. Simulations carried out in the Gazebo environment are used to validate the proposed active system.

This study investigates whether OpenAI's ChatGPT-3.5 and ChatGPT-4 can forecast future events. To evaluate the accuracy of the predictions, we take advantage of the fact that the training data at the time of our experiments (mid 2023) stopped at September 2021, and ask about events that happened in 2022. We employed two prompting strategies: direct prediction and what we call future narratives which ask ChatGPT to tell fictional stories set in the future with characters retelling events that happened in the past, but after ChatGPT's training data had been collected. We prompted ChatGPT to engage in storytelling, particularly within economic contexts. After analyzing 100 trials, we find that future narrative prompts significantly enhanced ChatGPT-4's forecasting accuracy. This was especially evident in its predictions of major Academy Award winners as well as economic trends, the latter inferred from scenarios where the model impersonated public figures like the Federal Reserve Chair, Jerome Powell. As a falsification exercise, we repeated our experiments in May 2024 at which time the models included more recent training data. ChatGPT-4's accuracy significantly improved when the training window included the events being prompted for, achieving 100% accuracy in many instances. The poorer accuracy for events outside of the training window suggests that in the 2023 prediction experiments, ChatGPT-4 was forming predictions based solely on its training data. Narrative prompting also consistently outperformed direct prompting. These findings indicate that narrative prompts leverage the models' capacity for hallucinatory narrative construction, facilitating more effective data synthesis and extrapolation than straightforward predictions. Our research reveals new aspects of LLMs' predictive capabilities and suggests potential future applications in analytical contexts.

This paper summarizes the experiments and results of the HYBRINFOX team for the CheckThat! 2024 - Task 1 competition. We propose an approach enriching Language Models such as RoBERTa with embeddings produced by triples (subject ; predicate ; object) extracted from the text sentences. Our analysis of the developmental data shows that this method improves the performance of Language Models alone. On the evaluation data, its best performance was in English, where it achieved an F1 score of 71.1 and ranked 12th out of 27 candidates. On the other languages (Dutch and Arabic), it obtained more mixed results. Future research tracks are identified toward adapting this processing pipeline to more recent Large Language Models.