Generalizing control policies to novel embodiments remains a fundamental challenge in enabling scalable and transferable learning in robotics. While prior works have explored this in locomotion, a systematic study in the context of manipulation tasks remains limited, partly due to the lack of standardized benchmarks. In this paper, we introduce a benchmark for learning cross-embodiment manipulation, focusing on two foundational tasks-reach and push-across a diverse range of morphologies. The benchmark is designed to test generalization along three axes: interpolation (testing performance within a robot category that shares the same link structure), extrapolation (testing on a robot with a different link structure), and composition (testing on combinations of link structures). On the benchmark, we evaluate the ability of different RL policies to learn from multiple morphologies and to generalize to novel ones. Our study aims to answer whether morphology-aware training can outperform single-embodiment baselines, whether zero-shot generalization to unseen morphologies is feasible, and how consistently these patterns hold across different generalization regimes. The results highlight the current limitations of multi-embodiment learning and provide insights into how architectural and training design choices influence policy generalization.

Autoencoders based on Graph Neural Networks (GNNs) have garnered significant attention in recent years for their ability to extract informative latent representations, characterizing the structure of complex topologies, such as graphs. Despite the prevalence of Graph Autoencoders, there has been limited focus on developing and evaluating explainable neural-based graph generative models specifically designed for signed networks. To address this gap, we propose the Signed Graph Archetypal Autoencoder (SGAAE) framework. SGAAE extracts node-level representations that express node memberships over distinct extreme profiles, referred to as archetypes, within the network. This is achieved by projecting the graph onto a learned polytope, which governs its polarization. The framework employs a recently proposed likelihood for analyzing signed networks based on the Skellam distribution, combined with relational archetypal analysis and GNNs. Our experimental evaluation demonstrates the SGAAEs' capability to successfully infer node memberships over the different underlying latent structures while extracting competing communities formed through the participation of the opposing views in the network. Additionally, we introduce the 2-level network polarization problem and show how SGAAE is able to characterize such a setting. The proposed model achieves high performance in different tasks of signed link prediction across four real-world datasets, outperforming several baseline models.

This paper proposes a new generative model and associated link inference method based on both node popularity and similarity. The starting point for the model is the prior work in [11] where the assortative mixed-membership stochastic blockmodel (AMMSB) was presented. In the prior model, link structure is generated via community strength (via a blockmodel) and community membership. In the new work, link structure is generated by using the prior model and adding "popularity" to the generative model. After the model is presented, the authors then derive an optimization criterion based upon a variational method (since exact inference is impossible).

Semantic relatedness between words has been extracted from a variety of sources.In this ongoing work, we explore and compare several options for determining if semantic relatedness can be extracted from navigation structures in Wikipedia. In that direction, we first investigate the potential of representation learning techniques such as DeepWalk in comparison to previously applied methods based on counting co-occurrences. Since both methods are based on (random) paths in the network, we also study different approaches to generate paths from Wikipedia link structure. For this task, we do not only consider the link structure of Wikipedia, but also actual navigation behavior of users. Finally, we analyze if semantics can also be extracted from smaller subsets of the Wikipedia link network. As a result we find that representation learning techniques mostly outperform the investigated co-occurrence counting methods on the Wikipedia network. However, we find that this is not the case for paths sampled from human navigation behavior.

Graph representation learning has achieved great success in many areas, including e-commerce, chemistry, biology, etc. However, the fundamental problem of choosing the appropriate dimension of node embedding for a given graph still remains unsolved. The commonly used strategies for Node Embedding Dimension Selection (NEDS) based on grid search or empirical knowledge suffer from heavy computation and poor model performance. In this paper, we revisit NEDS from the perspective of minimum entropy principle. Subsequently, we propose a novel Minimum Graph Entropy (MinGE) algorithm for NEDS with graph data. To be specific, MinGE considers both feature entropy and structure entropy on graphs, which are carefully designed according to the characteristics of the rich information in them. The feature entropy, which assumes the embeddings of adjacent nodes to be more similar, connects node features and link topology on graphs. The structure entropy takes the normalized degree as basic unit to further measure the higher-order structure of graphs. Based on them, we design MinGE to directly calculate the ideal node embedding dimension for any graph. Finally, comprehensive experiments with popular Graph Neural Networks (GNNs) on benchmark datasets demonstrate the effectiveness and generalizability of our proposed MinGE.

Bing announced a new link penalties. These link penalties are focused on taking down private blog networks (PBNs), subdomain leasing and manipulative cross-site linking. An inorganic site structure is a linking pattern that uses internal site-level link signals (with subdomains) or cross-site linking patterns (with external domains) in order to manipulate search engine rankings. While these spam techniques already existed, Bing introduced the concept of calling them "inorganic site structure" in order to describe them. Bing noted that sites legitimately create subdomains to keep different parts of the site separate, such as support.example.com.

With the huge amount of information inflowing the World Wide Web every second, it becomes more difficult and more difficult to retrieve information from the Web. This explains why the existence of a search engine is as important as the existence of the web itself. Since the appearance of the web, there has been a fundamental talk in the web research communit y to develop the rapid, effective, and precise search engines. This paper will be chiefly discussing about the most common search engine nowadays which is Google. The mathematical theory behind the Google search engine is the PageRank algorithm, which was presented by Sergey Brin and Lawrence Page [1].

Link prediction is a fundamental task in such areas as social network analysis, information retrieval, and bioinformatics. Usually link prediction methods use the link structures or node attributes as the sources of information. Recently, the relational topic model (RTM) and its variants have been proposed as hybrid methods that jointly model both sources of information and achieve very promising accuracy. However, the representations (features) learned by them are still not effective enough to represent the nodes (items). To address this problem, we generalize recent advances in deep learning from solely modeling i.i.d. sequences of attributes to jointly modeling graphs and non-i.i.d. sequences of attributes. Specifically, we follow the Bayesian deep learning framework and devise a hierarchical Bayesian model, called relational deep learning (RDL), to jointly model high-dimensional node attributes and link structures with layers of latent variables. Due to the multiple nonlinear transformations in RDL, standard variational inference is not applicable. We propose to utilize the product of Gaussians (PoG) structure in RDL to relate the inferences on different variables and derive a generalized variational inference algorithm for learning the variables and predicting the links. Experiments on three real-world datasets show that RDL works surprisingly well and significantly outperforms the state of the art.

In the past decade, social and information networks have become prevalent, and research on the network data has attracted much attention. Besides the link structure, network data are often equipped with the content information (i.e, node attributes) that is usually noisy and characterized by high dimensionality. As the curse of dimensionality could hamper the performance of many machine learning tasks on networks (e.g., community detection and link prediction), feature selection can be a useful technique for alleviating such issue. In this paper, we investigate the problem of unsupervised feature selection on networks. Most existing feature selection methods fail to incorporate the linkage information, and the state-of-the-art approaches usually rely on pseudo labels generated from clustering. Such cluster labels may be far from accurate and can mislead the feature selection process. To address these issues, we propose a generative point of view for unsupervised features selection on networks that can seamlessly exploit the linkage and content information in a more effective manner. We assume that the link structures and node content are generated from a succinct set of high-quality features, and we find these features through maximizing the likelihood of the generation process. Experimental results on three real-world datasets show that our approach can select more discriminative features than state-of-the-art methods.

Ensemble classification methods have been shown to produce more accurate predictions than the base component models. Due to their effectiveness, ensemble approaches have been applied in a wide range of domains to improve classification. The expected prediction error of classification models can be decomposed into bias and variance. Ensemble methods that independently construct component models (e.g., bagging) can improve performance by reducing the error due to variance, while methods that dependently construct component models (e.g., boosting) can improve performance by reducing the error due to bias and variance. Although ensemble methods were initially developed for classification of independent and identically distributed (i.i.d.) data, they can be directly applied for relational data by using a relational classifier as the base component model. This straightforward approach can improve classification for network data, but suffers from a number of limitations. First, relational data characteristics will only be exploited by the base relational classifier, and not by the ensemble algorithm itself. We note that explicitly accounting for the structured nature of relational data by the ensemble mechanism can significantly improve ensemble classification. Second, ensemble learning methods that assume i.i.d. data can fail to preserve the relational structure of non-i.i.d. data, which will (1) prevent the relational base classifiers from exploiting these structures, and (2) fail to accurately capture properties of the dataset, which can lead to inaccurate models and classifications. Third, ensemble mechanisms that assume i.i.d. data are limited to reducing errors associated with i.i.d. models and fail to reduce additional sources of error associated with more powerful (e.g., collective classification models. Our key observation is that collective classification methods have error due to variance in inference. This has been overlooked by current ensemble methods that assume exact inference methods and only focus on the typical goal of reducing errors due to learning, even if the methods explicitly consider relational data. Here we study the problem of ensemble classification for relational domains by focusing on the reduction of error due to variance. We propose a relational ensemble framework that explicitly accounts for the structured nature of relational data during both learning and inference. Our proposed framework consists of two components. (1) A method for learning accurate ensembles from relational data, focusing on the reduction of error due to variance in learning, while preserving the relational characteristics in the data. (2) A method for applying ensembles in collective classification contexts, focusing on further reduction of the error due to variance in inference, which has not been considered in state of the art ensemble methods.