TreeV AE hierarchically divides samples according to their intrinsic characteristics, shedding light on hidden structures in the data.

Results from experiments on three large real-world benchmark datasets demonstrate that OptFeature performs well in terms of accuracy and efficiency. Additional studies support the feasibility of our method.

Ubisoft, the video games publisher behind the Assassin's Creed series, has cancelled projects and announced a restructuring that will close several studios as a result of several years of weak results and disappointing sales. Ubisoft, the video games publisher behind the Assassin's Creed series, has cancelled projects and announced a restructuring that will close several studios as a result of several years of weak results and disappointing sales. The video game publisher behind the Assassin's Creed series has cancelled six projects including a remake of Prince of Persia: The Sands of Time as it fights to stay competitive in the global gaming market. Ubisoft announced a sweeping reorganisation and said it would cancel six games, sending its shares to their lowest level in more than a decade on Thursday. Ubisoft is abandoning development of six titles, including a highly anticipated remake of Prince of Persia - a series that dates back to 1989 and received an ill-fated Hollywood adaptation in 2010 - and delaying a further seven. Studios in Halifax, Canada and Stockholm are being closed, with restructuring to follow in other countries, it said.

Ubisoft has cancelled six video games - including its long-awaited Prince of Persia: The Sands of Time remake - as part of a major reset of its operations. The French developer and publisher, known for popular games such as Assassin's Creed, Far Cry and Just Dance, has closed two studios and delayed seven titles as part of its changes. Ubisoft boss Yves Guillemot said the move would create the conditions for a return to sustainable growth. The firm's shares plunged by 33% on Thursday morning following the announcement. The move comes at a time when studios are increasingly turning to video game remakes and remasters, with new versions of Super Mario Galaxy, Oblivion and Metal Gear Solid 3 proving popular in 2025.

Structure-inducing Language Models (SiLM) are trained on a self-supervised language modeling task, and induce a hierarchical sentence representation as a byproduct when processing an input. SiLMs couple strong syntactic generalization behavior with competitive performance on various NLP tasks, but many of their basic properties are yet underexplored. In this work, we train three different SiLM architectures from scratch: Structformer (Shen et al., 2021), UDGN (Shen et al., 2022), and GPST (Hu et al., 2024b). We train these architectures on both natural language (English, German, and Chinese) corpora and synthetic bracketing expressions. The models are then evaluated with respect to (i) properties of the induced syntactic representations (ii) performance on grammaticality judgment tasks, and (iii) training dynamics. We find that none of the three architectures dominates across all evaluation metrics. However, there are significant differences, in particular with respect to the induced syntactic representations. The Generative Pretrained Structured Transformer (GPST; Hu et al. 2024) performs most consistently across evaluation settings, and outperforms the other models on long-distance dependencies in bracketing expressions. Furthermore, our study shows that small models trained on large amounts of synthetic data provide a useful testbed for evaluating basic model properties.

Inspired by the success of language models (LM), scaling up deep learning recommendation systems (DLRS) has become a recent trend in the community. All previous methods tend to scale up the model parameters during training time. However, how to efficiently utilize and scale up computational resources during test time remains underexplored, which can prove to be a scaling-efficient approach and bring orthogonal improvements in LM domains. The key point in applying test-time scaling to DLRS lies in effectively generating diverse yet meaningful outputs for the same instance. We propose two ways: One is to explore the heterogeneity of different model architectures. The other is to utilize the randomness of model initialization under a homogeneous architecture. The evaluation is conducted across eight models, including both classic and SOTA models, on three benchmarks. Sufficient evidence proves the effectiveness of both solutions. We further prove that under the same inference budget, test-time scaling can outperform parameter scaling. Our test-time scaling can also be seamlessly accelerated with the increase in parallel servers when deployed online, without affecting the inference time on the user side. Code is available.

An associative memory (AM) enables cue-response recall, and it has recently been recognized as a key mechanism underlying modern neural architectures such as Transformers. In this work, we introduce the concept of distributed dynamic associative memory (DDAM), which extends classical AM to settings with multiple agents and time-varying data streams. In DDAM, each agent maintains a local AM that must not only store its own associations but also selectively memorize information from other agents based on a specified interest matrix. To address this problem, we propose a novel tree-based distributed online gradient descent algorithm, termed DDAM-TOGD, which enables each agent to update its memory on the fly via inter-agent communication over designated routing trees. We derive rigorous performance guarantees for DDAM-TOGD, proving sublinear static regret in stationary environments and a path-length dependent dynamic regret bound in non-stationary environments. These theoretical results provide insights into how communication delays and network structure impact performance. Building on the regret analysis, we further introduce a combinatorial tree design strategy that optimizes the routing trees to minimize communication delays, thereby improving regret bounds. Numerical experiments demonstrate that the proposed DDAM-TOGD framework achieves superior accuracy and robustness compared to representative online learning baselines such as consensus-based distributed optimization, confirming the benefits of the proposed approach in dynamic, distributed environments.