Generative modeling of graphs with spatial structure is essential across many applications from computer graphics to spatial genomics. Recent flow-based generative models have achieved impressive results by gradually adding and then learning to remove noise from these graphs. Existing models, however, use graph neural network architectures that are independent of the noise level, limiting their expressiveness. To address this issue, we introduce \textit{Noise-Conditioned Graph Networks} (NCGNs), a class of graph neural networks that dynamically modify their architecture according to the noise level during generation. Our theoretical and empirical analysis reveals that as noise increases, (1) graphs require information from increasingly distant neighbors and (2) graphs can be effectively represented at lower resolutions. Based on these insights, we develop Dynamic Message Passing (DMP), a specific instantiation of NCGNs that adapts both the range and resolution of message passing to the noise level. DMP consistently outperforms noise-independent architectures on a variety of domains including $3$D point clouds, spatiotemporal transcriptomics, and images. Code is available at https://github.com/peterpaohuang/ncgn.

The remarkable representational power of Vision Transformers (ViTs) remains underutilized in few-shot image classification. In this work, we introduce ViT-ProtoNet, which integrates a ViT-Small backbone into the Prototypical Network framework. By averaging class conditional token embeddings from a handful of support examples, ViT-ProtoNet constructs robust prototypes that generalize to novel categories under 5-shot settings. We conduct an extensive empirical evaluation on four standard benchmarks: Mini-ImageNet, FC100, CUB-200, and CIFAR-FS, including overlapped support variants to assess robustness. Across all splits, ViT-ProtoNet consistently outperforms CNN-based prototypical counterparts, achieving up to a 3.2\% improvement in 5-shot accuracy and demonstrating superior feature separability in latent space. Furthermore, it outperforms or is competitive with transformer-based competitors using a more lightweight backbone. Comprehensive ablations examine the impact of transformer depth, patch size, and fine-tuning strategy. To foster reproducibility, we release code and pretrained weights. Our results establish ViT-ProtoNet as a powerful, flexible approach for few-shot classification and set a new baseline for transformer-based meta-learners.

Continual learning aims to enable models to learn sequentially from continuously incoming data while retaining performance on previously learned tasks. With the Contrastive Language-Image Pre-trained model (CLIP) exhibiting strong capabilities across various downstream tasks, there has been growing interest in leveraging CLIP for continual learning in such scenarios. Most existing works overlook the inherent modality gap in CLIP, a key factor in its generalization and adaptability. In this paper, we analyze the variations in the modality gap during the fine-tuning of vision-language pre-trained models. Our observations reveal that the modality gap effectively reflects the extent to which pre-trained knowledge is preserved. Based on these insights, we propose a simple yet effective method, MG-CLIP, that improves CLIP's performance in class-incremental learning. Our approach leverages modality gap preservation to mitigate forgetting and modality gap compensation to enhance the capacity for new data, introducing a novel modality-gap-based perspective for continual learning. Extensive experiments on multiple benchmarks demonstrate that our method outperforms existing approaches without requiring additional replay data. Our code is available at https://github.

A rapidly alternating red and black display known as Ganzflicker induces visual hallucinations that reflect the generative capacity of the visual system. Recent proposals regarding the imagery spectrum, that is, differences in the visual system of individuals with absent imagery, typical imagery, and vivid imagery, suggest these differences should impact the complexity of other internally generated visual experiences. Here, we used tools from natural language processing to analyze free-text descriptions of hallucinations from over 4,000 participants, asking whether people with different imagery phenotypes see different things in their mind's eye during Ganzflicker-induced hallucinations. Strong imagers described complex, naturalistic content, while weak imagers reported simple geometric patterns. Embeddings from vision language models better captured these differences than text-only language models, and participants with stronger imagery used language with richer sensorimotor associations. These findings may reflect individual variation in coordination between early visual areas and higher-order regions relevant for the imagery spectrum.

Introduction: Electronic health records (EHR) are a critical medium through which patient stigmatization is perpetuated among healthcare teams. Methods: We identified linguistic features of doubt markers and stigmatizing labels in MIMIC-III EHR via expanded lexicon matching and supervised learning classifiers. Predictors of rates of linguistic features were assessed using Poisson regression models. Results: We found higher rates of stigmatizing labels per chart among patients who were Black or African American (RR: 1.16), patients with Medicare/Medicaid or government-run insurance (RR: 2.46), self-pay (RR: 2.12), and patients with a variety of stigmatizing disease and mental health conditions. Patterns among doubt markers were similar, though male patients had higher rates of doubt markers (RR: 1.25). We found increased stigmatizing labels used by nurses (RR: 1.40), and social workers (RR: 2.25), with similar patterns of doubt markers. Discussion: Stigmatizing language occurred at higher rates among historically stigmatized patients, perpetuated by multiple provider types.

We explore the use of Hamiltonian Monte Carlo (HMC) sampling as a probabilistic last layer approach for deep neural networks (DNNs). While HMC is widely regarded as a gold standard for uncertainty estimation, the computational demands limit its application to large-scale datasets and large DNN architectures. Although the predictions from the sampled DNN parameters can be parallelized, the computational cost still scales linearly with the number of samples (similar to an ensemble). Last layer HMC (LL--HMC) reduces the required computations by restricting the HMC sampling to the final layer of a DNN, making it applicable to more data-intensive scenarios with limited computational resources. In this paper, we compare LL-HMC against five last layer probabilistic deep learning (LL-PDL) methods across three real-world video datasets for driver action and intention. We evaluate the in-distribution classification performance, calibration, and out-of-distribution (OOD) detection. Due to the stochastic nature of the probabilistic evaluations, we performed five grid searches for different random seeds to avoid being reliant on a single initialization for the hyperparameter configurations. The results show that LL--HMC achieves competitive in-distribution classification and OOD detection performance. Additional sampled last layer parameters do not improve the classification performance, but can improve the OOD detection. Multiple chains or starting positions did not yield consistent improvements.

Recent studies have proposed interpreting the training process from an ergodic perspective. Building on this foundation, we present a unified framework for understanding and accelerating the training of deep neural networks via stochastic gradient descent (SGD). By analyzing the geometric landscape of the objective function we introduce a practical diagnostic, the running estimate of the largest Lyapunov exponent, which provably distinguishes genuine convergence toward stable minimizers from mere statistical stabilization near saddle points. We then propose a ghost category extension for standard classifiers that adds auxiliary ghost output nodes so the model gains extra descent directions that open a lateral corridor around narrow loss barriers and enable the optimizer to bypass poor basins during the early training phase. We show that this extension strictly reduces the approximation error and that after sufficient convergence the ghost dimensions collapse so that the extended model coincides with the original one and there exists a path in the enlarged parameter space along which the total loss does not increase. Taken together, these results provide a principled architecture level intervention that accelerates early stage trainability while preserving asymptotic behavior and simultaneously serves as an architecture-friendly regularizer.

Can machines truly think, reason and act in domains like humans? This enduring question continues to shape the pursuit of Artificial General Intelligence (AGI). Despite the growing capabilities of models such as GPT-4.5, DeepSeek, Claude 3.5 Sonnet, Phi-4, and Grok 3, which exhibit multimodal fluency and partial reasoning, these systems remain fundamentally limited by their reliance on token-level prediction and lack of grounded agency. This paper offers a cross-disciplinary synthesis of AGI development, spanning artificial intelligence, cognitive neuroscience, psychology, generative models, and agent-based systems. We analyze the architectural and cognitive foundations of general intelligence, highlighting the role of modular reasoning, persistent memory, and multi-agent coordination. In particular, we emphasize the rise of Agentic RAG frameworks that combine retrieval, planning, and dynamic tool use to enable more adaptive behavior. We discuss generalization strategies, including information compression, test-time adaptation, and training-free methods, as critical pathways toward flexible, domain-agnostic intelligence. Vision-Language Models (VLMs) are reexamined not just as perception modules but as evolving interfaces for embodied understanding and collaborative task completion. We also argue that true intelligence arises not from scale alone but from the integration of memory and reasoning: an orchestration of modular, interactive, and self-improving components where compression enables adaptive behavior. Drawing on advances in neurosymbolic systems, reinforcement learning, and cognitive scaffolding, we explore how recent architectures begin to bridge the gap between statistical learning and goal-directed cognition. Finally, we identify key scientific, technical, and ethical challenges on the path to AGI.

Estimating the score function (or other population-density-dependent functions) is a fundamental component of most generative models. However, such function estimation is computationally and statistically challenging. Can we avoid function estimation for data generation? We propose an estimation-free generative method: A set of points whose locations are deterministically updated with (inverse) gradient descent can transport a uniform distribution to arbitrary data distribution, in the mean field regime, without function estimation, training neural networks, and even noise injection. The proposed method is built upon recent advances in the physics of interacting particles. We show, both theoretically and experimentally, that these advances can be leveraged to develop novel generative methods.

Grokking refers to a delayed generalization following overfitting when optimizing artificial neural networks with gradient-based methods. In this work, we demonstrate that grokking can be induced by regularization, either explicit or implicit. More precisely, we show that when there exists a model with a property $P$ (e.g., sparse or low-rank weights) that generalizes on the problem of interest, gradient descent with a small but non-zero regularization of $P$ (e.g., $\ell_1$ or nuclear norm regularization) results in grokking. This extends previous work showing that small non-zero weight decay induces grokking. Moreover, our analysis shows that over-parameterization by adding depth makes it possible to grok or ungrok without explicitly using regularization, which is impossible in shallow cases. We further show that the $\ell_2$ norm is not a reliable proxy for generalization when the model is regularized toward a different property $P$, as the $\ell_2$ norm grows in many cases where no weight decay is used, but the model generalizes anyway. We also show that grokking can be amplified solely through data selection, with any other hyperparameter fixed.