Large Language Models (LLMS) have increasingly become central to generating content with potential societal impacts. Notably, these models have demonstrated capabilities for generating content that could be deemed harmful. To mitigate these risks, researchers have adopted safety training techniques to align model outputs with societal values to curb the generation of malicious content. However, the phenomenon of "jailbreaking", where carefully crafted prompts elicit harmful responses from models, persists as a significant challenge. This research conducts a comprehensive analysis of existing studies on jailbreaking LLMs and their defense techniques. We meticulously investigate nine attack techniques and seven defense techniques applied across three distinct language models: Vicuna, LLama, and GPT-3.5 Turbo. We aim to evaluate the effectiveness of these attack and defense techniques. Our findings reveal that existing white-box attacks underperform compared to universal techniques and that including special tokens in the input significantly affects the likelihood of successful attacks. This research highlights the need to concentrate on the security facets of LLMs. Additionally, we contribute to the field by releasing our datasets and testing framework, aiming to foster further research into LLM security. We believe these contributions will facilitate the exploration of security measures within this domain.

Deep learning models have performed well on many NLP tasks. However, their internal mechanisms are typically difficult for humans to understand. The development of methods to explain models has become a key issue in the reliability of deep learning models in many important applications. Various saliency explanation methods, which give each feature of input a score proportional to the contribution of output, have been proposed to determine the part of the input which a model values most. Despite a considerable body of work on the evaluation of saliency methods, whether the results of various evaluation metrics agree with human cognition remains an open question. In this study, we propose a new human-based method to evaluate saliency methods in NLP by crowdsourcing. We recruited 800 crowd workers and empirically evaluated seven saliency methods on two datasets with the proposed method. We analyzed the performance of saliency methods, compared our results with existing automated evaluation methods, and identified notable differences between NLP and computer vision (CV) fields when using saliency methods. The instance-level data of our crowdsourced experiments and the code to reproduce the explanations are available at https://github.com/xtlu/lreccoling_evaluation.

Abstract--Graph Machine Learning (Graph ML) has witnessed substantial advancements in recent years. With their remarkable ability to process graph-structured data, Graph ML techniques have been extensively utilized across diverse applications, including critical domains like finance, healthcare, and transportation. Despite their societal benefits, recent research highlights significant safety concerns associated with the widespread use of Graph ML models. Lacking safety-focused designs, these models can produce unreliable predictions, demonstrate poor generalizability, and compromise data confidentiality. In high-stakes scenarios such as financial fraud detection, these vulnerabilities could jeopardize both individuals and society at large. Therefore, it is imperative to prioritize the development of safety-oriented Graph ML models to mitigate these risks and enhance public confidence in their applications. In this survey paper, we explore three critical aspects vital for enhancing safety in Graph ML: reliability, generalizability, and confidentiality. We categorize and analyze threats to each aspect under three headings: model threats, data threats, and attack threats. This novel taxonomy guides our review of effective strategies to protect against these threats. Our systematic review lays a groundwork for future research aimed at developing practical, safety-centered Graph ML models. Furthermore, we highlight the significance of safe Graph ML practices and suggest promising avenues for further investigation in this crucial area. To prevalent across a wide range of real-world applications, narrow this gap, our survey seeks to resolve two critical including drug discovery [15], traffic forecasting questions: (1) What are the key aspects involved in the safety [76], and disease diagnosis [96]. Within these domains, issues of Graph ML? (2) What specific types of threats might Graph Machine Learning (Graph ML) plays a pivotal role in arise within each aspect, and how can they be effectively modeling this data and executing graph-based predictive handled? To address the first question, we introduce a novel tasks [83], [187]. However, as the scope of Graph ML taxonomy that facilitates a thorough categorization of safety applications expands, concerns about their underlying safety issues in Graph ML. To answer the second question, we issues intensify [37].

We discuss the role of humans in algorithmic decision-making (ADM) for socially relevant problems from a technical and philosophical perspective. In particular, we illustrate tensions arising from diverse expectations, values, and constraints by and on the humans involved. To this end, we assume that a strategic decision-maker (SDM) introduces ADM to optimize strategic and societal goals while the algorithms' recommended actions are overseen by a practical decision-maker (PDM) - a specific human-in-the-loop - who makes the final decisions. While the PDM is typically assumed to be a corrective, it can counteract the realization of the SDM's desired goals and societal values not least because of a misalignment of these values and unmet information needs of the PDM. This has significant implications for the distribution of power between the stakeholders in ADM, their constraints, and information needs. In particular, we emphasize the overseeing PDM's role as a potential political and ethical decision maker, who acts expected to balance strategic, value-driven objectives and on-the-ground individual decisions and constraints. We demonstrate empirically, on a machine learning benchmark dataset, the significant impact an overseeing PDM's decisions can have even if the PDM is constrained to performing only a limited amount of actions differing from the algorithms' recommendations. To ensure that the SDM's intended values are realized, the PDM needs to be provided with appropriate information conveyed through tailored explanations and its role must be characterized clearly. Our findings emphasize the need for an in-depth discussion of the role and power of the PDM and challenge the often-taken view that just including a human-in-the-loop in ADM ensures the 'correct' and 'ethical' functioning of the system.

One can improve predictability in the unknown domain by combining forecasts of imperfect complex computational models using a Bayesian statistical machine learning framework. In many cases, however, the models used in the mixing process are similar. In addition to contaminating the model space, the existence of such similar, or even redundant, models during the multimodeling process can result in misinterpretation of results and deterioration of predictive performance. In this work we describe a method based on the Principal Component Analysis that eliminates model redundancy. We show that by adding model orthogonalization to the proposed Bayesian Model Combination framework, one can arrive at better prediction accuracy and reach excellent uncertainty quantification performance.

Only 16 months have passed, but the release of ChatGPT back in November 2022 feels already like ancient AI history. Hundreds of billions of dollars, both public and private, are pouring into AI. Thousands of AI-powered products have been created, including the new GPT-4o just this week. Everyone from students to scientists now use these large language models. Our world, and in particular the world of AI, has decidedly changed.

Machine learning analysis of data gathered by acoustic recording devices is a promising new tool for monitoring the marbled murrelet and other secretive, hard-to-study species, research by Oregon State University and the U.S. Forest Service has shown. The threatened marbled murrelet is an iconic Pacific Northwest seabird that's closely related to puffins and murres, but unlike those birds, murrelets raise their young as far as 60 miles inland in mature and old-growth forests. "There are very few species like it," said co-author Matt Betts of the OSU College of Forestry. "And there's no other bird that feeds in the ocean and travels such long distances to inland nest sites. This behavior is super unusual and it makes studying this bird really challenging."

Federated Learning (FL) has emerged as a leading paradigm for decentralized, privacy preserving machine learning training. However, recent research on gradient inversion attacks (GIAs) have shown that gradient updates in FL can leak information on private training samples. While existing surveys on GIAs have focused on the honest-but-curious server threat model, there is a dearth of research categorizing attacks under the realistic and far more privacy-infringing cases of malicious servers and clients. In this paper, we present a survey and novel taxonomy of GIAs that emphasize FL threat models, particularly that of malicious servers and clients. We first formally define GIAs and contrast conventional attacks with the malicious attacker. We then summarize existing honest-but-curious attack strategies, corresponding defenses, and evaluation metrics. Critically, we dive into attacks with malicious servers and clients to highlight how they break existing FL defenses, focusing specifically on reconstruction methods, target model architectures, target data, and evaluation metrics. Lastly, we discuss open problems and future research directions.

Federated Learning (FL) has become an attractive approach to collaboratively train Machine Learning (ML) models while data sources' privacy is still preserved. However, most of existing FL approaches are based on supervised techniques, which could require resource-intensive activities and human intervention to obtain labelled datasets. Furthermore, in the scope of cyberattack detection, such techniques are not able to identify previously unknown threats. In this direction, this work proposes a novel unsupervised FL approach for the identification of potential misbehavior in vehicular environments. We leverage the computing capabilities of public cloud services for model aggregation purposes, and also as a central repository of misbehavior events, enabling cross-vehicle learning and collective defense strategies. Our solution integrates the use of Gaussian Mixture Models (GMM) and Variational Autoencoders (VAE) on the VeReMi dataset in a federated environment, where each vehicle is intended to train only with its own data. Furthermore, we use Restricted Boltzmann Machines (RBM) for pre-training purposes, and Fedplus as aggregation function to enhance model's convergence. Our approach provides better performance (more than 80 percent) compared to recent proposals, which are usually based on supervised techniques and artificial divisions of the VeReMi dataset.

Budgetary constraints force organizations to pursue only a subset of possible innovation projects. Identifying which subset is most promising is an error-prone exercise, and involving multiple decision makers may be prudent. This raises the question of how to most effectively aggregate their collective nous. Our model of organizational portfolio selection provides some first answers. We show that portfolio performance can vary widely. Delegating evaluation makes sense when organizations employ the relevant experts and can assign projects to them. In most other settings, aggregating the impressions of multiple agents leads to better performance than delegation. In particular, letting agents rank projects often outperforms alternative aggregation rules -- including averaging agents' project scores as well as counting their approval votes -- especially when organizations have tight budgets and can select only a few project alternatives out of many.