Link prediction, which aims to forecast unseen connections in graphs, is a fundamental task in graph machine learning. Heuristic methods, leveraging a range of different pairwise measures such as common neighbors and shortest paths, often rival the performance of vanilla Graph Neural Networks (GNNs). Therefore, recent advancements in GNNs for link prediction (GNN4LP) have primarily focused on integrating one or a few types of pairwise information. In this work, we reveal that different node pairs within the same dataset necessitate varied pairwise information for accurate prediction and models that only apply the same pairwise information uniformly could achieve suboptimal performance. As a result, we propose a simple mixture of experts model Link-MoE for link prediction. Link-MoE utilizes various GNNs as experts and strategically selects the appropriate expert for each node pair based on various types of pairwise information. Experimental results across diverse real-world datasets demonstrate substantial performance improvement from Link-MoE. Notably, Link-MoE achieves a relative improvement of 18.71% on the MRR metric for the Pubmed dataset and 9.59% on the Hits@100 metric for the ogbl-ppa dataset, compared to the best baselines. The code is available at https://github.com/ml-ml/Link-MoE/.

The category gap between training and evaluation has been characterised as one of the main obstacles to the success of Few-Shot Learning (FSL). In this paper, we for the first time empirically identify image background, common in realistic images, as a shortcut knowledge helpful for in-class classification but ungeneralizable beyond training categories in FSL. A novel framework, COSOC, is designed to tackle this problem by extracting foreground objects in images at both training and evaluation without any extra supervision. Extensive experiments carried on inductive FSL tasks demonstrate the effectiveness of our approaches.

To solve the variational problem in Equation (A.5), we can define a factorized variational family Returning to the variational problem in Equation (A.5), we can now write D Following Haarnoja et al. [16], we define the variational distribution over next states to be the the true transition dynamics, that is, q Corollary 1 (Fixed-Time Outcome-Driven Reward Function). Parts of this proof are adapted from the proof given in Haarnoja et al. [16], modified for the Bellman operator proposed in Definition 1. 1. Outcome-Driven Policy Evaluation (ODPE): Instead of absorbing the entropy term into the Q-function, we can define an entropy-augmented reward as r The first condition is true by assumption. Therefore, we apply convergence results for policy evaluation with transition-dependent discount factors [53] to this contraction mapping, and the result immediately follows. Convergence follows from Outcome-Driven Policy Evaluation above. Remark 2. The convergence proof of ODPE assumes a transition-dependent discount factor [53], because the variational distribution used in Equation (11) depends on the next state and action as well as on the desired outcome.

Figure 2: The figure presents still images extracted from 360 videos used in the experiment to display various environments to the participants. The videos were selected from the publically available 360 VR video dataset (Li et al. (2017).) The EEVR dataset comprises synchronized pairs of physiological signals and textual data. It includes responses to four self-assessment questions regarding perceived arousal, valence, dominance, and discrete emotions ratings collected using PANAS questionnaires (which were further utilized to calculate Positive and Negative Affect Score). The EEVR dataset was collected using Virtual Reality (VR) 360 videos as the elicitation medium. The videos utilized in the dataset were selected based on their arousal and valence ratings to cover all four quadrants of the Russell circumplex emotion model (Russell et al. (1989)), as shown in Figure 2. The remainder of the supplementary materials provide detailed information about the EEVR dataset. Figure 3 provides a datasheet for the EEVR dataset based on Gebru et al. (2018).

EEVR (Emotion Elicitation in Virtual Reality) is a novel dataset specifically designed for language supervision-based pre-training of emotion recognition tasks, such as valence and arousal classification. It features high-quality physiological signals, including electrodermal activity (EDA) and photoplethysmography (PPG), acquired through emotion elicitation via 360-degree virtual reality (VR) videos. Additionally, it includes subject-wise textual descriptions of emotions experienced during each stimulus gathered from qualitative interviews. The dataset consists of recordings from 37 participants and is the first dataset to pair raw text with physiological signals, providing additional contextual information that objective labels cannot offer. To leverage this dataset, we introduced the Contrastive Language Signal Pre-training (CLSP) method, which jointly learns representations using pairs of physiological signals and textual descriptions. Our results show that integrating self-reported textual descriptions with physiological signals significantly improves performance on emotion recognition tasks, such as arousal and valence classification. Moreover, our pre-trained CLSP model demonstrates strong zero-shot transferability to existing datasets, outperforming supervised baseline models, suggesting that the representations learned by our method are more contextualized and generalized. The dataset also includes baseline models for arousal, valence, and emotion classification, as well as code for data cleaning and feature extraction.

Gaze following and social gaze prediction are fundamental tasks providing insights into human communication behaviors, intent, and social interactions. Most previous approaches addressed these tasks separately, either by designing highly specialized social gaze models that do not generalize to other social gaze tasks or by considering social gaze inference as an ad-hoc post-processing of the gaze following task. Furthermore, the vast majority of gaze following approaches have proposed models that can handle only one person at a time and are static, therefore failing to take advantage of social interactions and temporal dynamics. In this paper, we address these limitations and introduce a novel framework to jointly predict the gaze target and social gaze label for all people in the scene. It comprises (i) a temporal, transformer-based architecture that, in addition to frame tokens, handles personspecific tokens capturing the gaze information related to each individual; (ii) a new dataset, VSGaze, built from multiple gaze following and social gaze datasets by extending and validating head detections and tracks, and unifying annotation types. We demonstrate that our model can address and benefit from training on all tasks jointly, achieving state-of-the-art results for multi-person gaze following and social gaze prediction. Our annotations and code will be made publicly available.

In this section, we present the proofs to the theorems introduced in the main paper. A.1 Proof to Theorem 1 Theorem 1 describes the concavity of objective of the selective rationalization (Equation (7)). The proof is presented as follows. As can be seen, the fundamental cause of the concavity is the overfitting nature of the rationale predictor. More specifically, having one predictor trained on multiple rationale selections is worse than having multiple predictors, each specializing in a single corner case.

Sign up for the Slatest to get the most insightful analysis, criticism, and advice out there, delivered to your inbox daily. Over the past week, at least two venerable American newspapers--the Chicago Sun-Times and the Philadelphia Inquirer--published a 56-page insert of summer content that was in large part produced by A.I. The most glaring evidence was a now-notorious "summer reading list," which recommended 15 books, five of them real, 10 of them imaginary, with summaries of fake titles like Isabel Allende's Tidewater Dreams, Min Jin Lee's Nightshade Market, Rebecca Makkai's Boiling Point, and Percival Everett's The Rainmakers. The authors exist; the books do not. The rest of the section, which included anodyne listicles about summer activities, barbecuing, and photography, soon attracted additional scrutiny.

Putting on Project Moohan, an upcoming XR headset developed by Google, Samsung, and Qualcomm, for the first time felt strangely familiar. From twisting the head-strap knob on the back to slipping the standalone battery pack into my pants pocket, my mind was transported back to February 2024, when I tried on the Apple Vision Pro on launch day. Also: The best smart glasses unveiled at I/O 2025 weren't made by Google Only this time, the headset was powered by Android XR, Google's newest operating system built around Gemini, the same AI model that dominated the Google I/O headlines this week. The difference in software was immediately noticeable -- from the home grid of Google apps like Photos, Maps, and YouTube (which VisionOS still lacks) to prompting for Gemini instead of Siri with a long press of the headset's multifunctional key. While my demo with Project Moohan lasted only about 10 minutes, it gave me a clear understanding of how it's challenging Apple's Vision Pro and how Google, Samsung, and Qualcomm plan to convince the masses that the future of spatial computing does, in fact, live in a bulkier space-helmet-like device. For starters, there's no denying that the industrial designers of Project Moohan drew some inspiration from the Apple Vision Pro.

At a company-wide hackathon this month, developers at finance firm Block built a dizzying number of prototype tools including a database debugger, a program for identifying duplicated code, and an app that automates Bitcoin support. The sudden productivity boost was driven by Goose, an artificial intelligence agent developed by Block several months ago that can help with coding and other work like knocking together data visualizations or mocking up new product features. "We've always had really strong hack weeks, but this one was at another level," says Jackie Brosamer, who leads the AI and data platform at Block. "We have tens of ideas that we're looking to bring to production." Goose helped developers at Block to develop a new agent-to-agent communication server at the hackathon.