With the increasing computation of training graph neural networks (GNNs) on large-scale graphs, graph condensation (GC) has emerged as a promising solution to synthesize a compact, substitute graph of the large-scale original graph for efficient GNN training. However, existing GC methods predominantly employ classification as the surrogate task for optimization, thus excessively relying on node labels and constraining their utility in label-sparsity scenarios. More critically, this surrogate task tends to overfit class-specific information within the condensed graph, consequently restricting the generalization capabilities of GC for other downstream tasks. To address these challenges, we introduce Contrastive Graph Condensation (CTGC), which adopts a self-supervised surrogate task to extract critical, causal information from the original graph and enhance the cross-task generalizability of the condensed graph. Specifically, CTGC employs a dual-branch framework to disentangle the generation of the node attributes and graph structures, where a dedicated structural branch is designed to explicitly encode geometric information through nodes' positional embeddings. By implementing an alternating optimization scheme with contrastive loss terms, CTGC promotes the mutual enhancement of both branches and facilitates high-quality graph generation through the model inversion technique. Extensive experiments demonstrate that CTGC excels in handling various downstream tasks with a limited number of labels, consistently outperforming state-of-the-art GC methods.

This paper introduces a novel XAI approach based on near-misses analysis (NMA). This approach reveals a hierarchy of logical 'concepts' inferred from the latent decision-making process of a Neural Network (NN) without delving into its explicit structure. We examined our proposed XAI approach on different network architectures that vary in size and shape (e.g., ResNet, VGG, EfficientNet, MobileNet) on several datasets (ImageNet and CIFAR100). The results demonstrate its usability to reflect NNs latent process of concepts generation. We generated a new metric for explainability. Moreover, our experiments suggest that efficient architectures, which achieve a similar accuracy level with much less neurons may still pay the price of explainability and robustness in terms of concepts generation. We, thus, pave a promising new path for XAI research to follow.

When we speak, write or listen, we continuously make predictions based on our knowledge of a language's grammar. Remarkably, children acquire this grammatical knowledge within just a few years, enabling them to understand and generalise to novel constructions that have never been uttered before. Language models are powerful tools that create representations of language by incrementally predicting the next word in a sentence, and they have had a tremendous societal impact in recent years. The central research question of this thesis is whether these models possess a deep understanding of grammatical structure similar to that of humans. This question lies at the intersection of natural language processing, linguistics, and interpretability. To address it, we will develop novel interpretability techniques that enhance our understanding of the complex nature of large-scale language models. We approach our research question from three directions. First, we explore the presence of abstract linguistic information through structural priming, a key paradigm in psycholinguistics for uncovering grammatical structure in human language processing. Next, we examine various linguistic phenomena, such as adjective order and negative polarity items, and connect a model's comprehension of these phenomena to the data distribution on which it was trained. Finally, we introduce a controlled testbed for studying hierarchical structure in language models using various synthetic languages of increasing complexity and examine the role of feature interactions in modelling this structure. Our findings offer a detailed account of the grammatical knowledge embedded in language model representations and provide several directions for investigating fundamental linguistic questions using computational methods.

Ultrasound imaging is widely used in clinical diagnosis due to its non-invasive nature and real-time capabilities. However, conventional ultrasound diagnostics face several limitations, including high dependence on physician expertise and suboptimal image quality, which complicates interpretation and increases the likelihood of diagnostic errors. Artificial intelligence (AI) has emerged as a promising solution to enhance clinical diagnosis, particularly in detecting abnormalities across various biomedical imaging modalities. Nonetheless, current AI models for ultrasound imaging face critical challenges. First, these models often require large volumes of labeled medical data, raising concerns over patient privacy breaches. Second, most existing models are task-specific, which restricts their broader clinical utility. To overcome these challenges, we present UltraFedFM, an innovative privacy-preserving ultrasound foundation model. UltraFedFM is collaboratively pre-trained using federated learning across 16 distributed medical institutions in 9 countries, leveraging a dataset of over 1 million ultrasound images covering 19 organs and 10 ultrasound modalities. This extensive and diverse data, combined with a secure training framework, enables UltraFedFM to exhibit strong generalization and diagnostic capabilities. It achieves an average area under the receiver operating characteristic curve of 0.927 for disease diagnosis and a dice similarity coefficient of 0.878 for lesion segmentation. Notably, UltraFedFM surpasses the diagnostic accuracy of mid-level ultrasonographers and matches the performance of expert-level sonographers in the joint diagnosis of 8 common systemic diseases. These findings indicate that UltraFedFM can significantly enhance clinical diagnostics while safeguarding patient privacy, marking an advancement in AI-driven ultrasound imaging for future clinical applications.

With the growing popularity of generative AI for images, video, and music, we witnessed models rapidly improve in quality and performance. However, not much attention is paid towards enabling AI's ability to "be creative". In this study, we implemented and simulated the systems model of creativity (proposed by Csikszentmihalyi) using virtual agents utilizing large language models (LLMs) and text prompts. For comparison, the simulations were conducted with the "virtual artists" being: 1)isolated and 2)placed in a multi-agent system. Both scenarios were compared by analyzing the variations and overall "creativity" in the generated artifacts (measured via a user study and LLM). Our results suggest that the generative agents may perform better in the framework of the systems model of creativity.

Training large language models (LLMs) to spend more time thinking and reflection before responding is crucial for effectively solving complex reasoning tasks in fields such as science, coding, and mathematics. However, the effectiveness of mechanisms like self-reflection and self-correction depends on the model's capacity to accurately assess its own performance, which can be limited by factors such as initial accuracy, question difficulty, and the lack of external feedback. In this paper, we delve into a two-player paradigm that separates the roles of reasoning and critique models, where the critique model provides step-level feedback to supervise the reasoning (actor) model during both test-time and training-time. We first propose AutoMathCritique, an automated and scalable framework for collecting critique data, resulting in a dataset of 76, 321 responses paired with step-level feedback. Fine-tuning language models with this dataset enables them to generate natural language feedback for mathematical reasoning. We demonstrate that the critique models consistently improve the actor's performance on difficult queries at test-time, especially when scaling up inference-time computation. Motivated by these findings, we introduce the critique-based supervision to the actor's selftraining process, and propose a critique-in-the-loop self-improvement method. Experiments show that the method improves the actor's exploration efficiency and solution diversity, especially on challenging queries, leading to a stronger reasoning model. Lastly, we take the preliminary step to explore training self-talk reasoning models via critique supervision and showcase their potential.

Jailbreak prompts pose a significant threat in AI and cybersecurity, as they are crafted to bypass ethical safeguards in large language models, potentially enabling misuse by cybercriminals. This paper analyzes jailbreak prompts from a cyber defense perspective, exploring techniques like prompt injection and context manipulation that allow harmful content generation, content filter evasion, and sensitive information extraction. We assess the impact of successful jailbreaks, from misinformation and automated social engineering to hazardous content creation, including bioweapons and explosives. To address these threats, we propose strategies involving advanced prompt analysis, dynamic safety protocols, and continuous model fine-tuning to strengthen AI resilience. Additionally, we highlight the need for collaboration among AI researchers, cybersecurity experts, and policymakers to set standards for protecting AI systems. Through case studies, we illustrate these cyber defense approaches, promoting responsible AI practices to maintain system integrity and public trust. \textbf{\color{red}Warning: This paper contains content which the reader may find offensive.}

Users may inadvertently upload personally identifiable information (PII) to Machine Learning as a Service (MLaaS) providers. When users no longer want their PII on these services, regulations like GDPR and COPPA mandate a right to forget for these users. As such, these services seek efficient methods to remove the influence of specific data points. Thus the introduction of machine unlearning. Traditionally, unlearning is performed with the removal of entire data samples (sample unlearning) or whole features across the dataset (feature unlearning). However, these approaches are not equipped to handle the more granular and challenging task of unlearning specific objects within a sample. To address this gap, we propose a scene graph-based object unlearning framework. This framework utilizes scene graphs, rich in semantic representation, transparently translate unlearning requests into actionable steps. The result, is the preservation of the overall semantic integrity of the generated image, bar the unlearned object. Further, we manage high computational overheads with influence functions to approximate the unlearning process. For validation, we evaluate the unlearned object's fidelity in outputs under the tasks of image reconstruction and image synthesis. Our proposed framework demonstrates improved object unlearning outcomes, with the preservation of unrequested samples in contrast to sample and feature learning methods. This work addresses critical privacy issues by increasing the granularity of targeted machine unlearning through forgetting specific object-level details without sacrificing the utility of the whole data sample or dataset feature.

Crop yield prediction is essential for agricultural planning but remains challenging due to the complex interactions between weather, climate, and management practices. To address these challenges, we introduce a deep learning-based multi-model called Climate-Management Aware Vision Transformer (CMAViT), designed for pixel-level vineyard yield predictions. CMAViT integrates both spatial and temporal data by leveraging remote sensing imagery and short-term meteorological data, capturing the effects of growing season variations. Additionally, it incorporates management practices, which are represented in text form, using a cross-attention encoder to model their interaction with time-series data. This innovative multi-modal transformer tested on a large dataset from 2016-2019 covering 2,200 hectares and eight grape cultivars including more than 5 million vines, outperforms traditional models like UNet-ConvLSTM, excelling in spatial variability capture and yield prediction, particularly for extreme values in vineyards. CMAViT achieved an R2 of 0.84 and a MAPE of 8.22% on an unseen test dataset. Masking specific modalities lowered performance: excluding management practices, climate data, and both reduced R2 to 0.73, 0.70, and 0.72, respectively, and raised MAPE to 11.92%, 12.66%, and 12.39%, highlighting each modality's importance for accurate yield prediction. Code is available at https://github.com/plant-ai-biophysics-lab/CMAViT.

Autonomous driving involves complex tasks such as data fusion, object and lane detection, behavior prediction, and path planning. As opposed to the modular approach which dedicates individual subsystems to tackle each of those tasks, the end-to-end approach treats the problem as a single learnable task using deep neural networks, reducing system complexity and minimizing dependency on heuristics. Conditional imitation learning (CIL) trains the end-to-end model to mimic a human expert considering the navigational commands guiding the vehicle to reach its destination, CIL adopts specialist network branches dedicated to learn the driving task for each navigational command. Nevertheless, the CIL model lacked generalization when deployed to unseen environments. This work introduces the conditional imitation co-learning (CIC) approach to address this issue by enabling the model to learn the relationships between CIL specialist branches via a co-learning matrix generated by gated hyperbolic tangent units (GTUs). Additionally, we propose posing the steering regression problem as classification, we use a classification-regression hybrid loss to bridge the gap between regression and classification, we also propose using co-existence probability to consider the spatial tendency between the steering classes. Our model is demonstrated to improve autonomous driving success rate in unseen environment by 62% on average compared to the CIL method.