Huawei Technologies (United Kingdom)

Continual learning (CL) aims to equip models with the ability to learn from a stream of tasks without forgetting previous knowledge. With the progress of vision-language models like Contrastive Language-Image Pre-training (CLIP), their promise for CL has attracted increasing attention due to their strong generalizability. However, the potential of rich textual semantic priors in CLIP in addressing the stability–plasticity dilemma remains underexplored. During backbone training, most approaches transfer past knowledge without considering semantic relevance, leading to interference from unrelated tasks that disrupt the balance between stability and plasticity. Besides, while text-based classifiers provide strong generalization, they suffer from limited plasticity due to the inherent modality gap in CLIP. Visual classifiers help bridge this gap, but their prototypes lack rich and precise semantics. To address these challenges, we propose Semantic-Enriched Continual Adaptation (SECA), a unified framework that harnesses the anti-forgetting and structured nature of textual priors to guide semantic-aware knowledge transfer in the backbone and reinforce the semantic structure of the visual classifier. Specifically, a Semantic-Guided Adaptive Knowledge Transfer (SG-AKT) module is proposed to assess new images' relevance to diverse historical visual knowledge via textual cues, and aggregate relevant knowledge in an instance-adaptive manner as distillation signals. Moreover, a Semantic-Enhanced Visual Prototype Refinement (SE-VPR) module is introduced to refine visual prototypes using inter-class semantic relations captured in class-wise textual embeddings. Extensive experiments on multiple benchmarks validate the effectiveness of our approach.

Multimodal Large Language Models (MLLMs) have played an increasingly important role in multimodal intelligence. However, the existing fine-tuning methods often ignore cross-modal heterogeneity, limiting their full potential. In this work, we propose a novel fine-tuning strategy by injecting beneficial random noise, which outperforms previous methods and even surpasses full fine-tuning, with minimal additional parameters. The proposed Multimodal Noise Generator (MuNG) enables efficient modality fine-tuning by injecting customized noise into the frozen MLLMs. Specifically, we reformulate the reasoning process of MLLMs from a variational inference perspective, upon which we design a multimodal noise generator that dynamically analyzes cross-modal relationships in image-text pairs to generate task adaptive beneficial noise. Injecting this type of noise into the MLLMs effectively suppresses irrelevant semantic components, leading to significantly improved cross-modal representation alignment and enhanced performance on downstream tasks. Experiments on two mainstream MLLMs, QwenVL and LLaVA, demonstrate that our method surpasses full parameter fine-tuning and other existing fine-tuning approaches, while requiring adjustments to only about 1~2% additional parameters.

Retrieval-Augmented Generation (RAG) has proven effective in integrating external knowledge into large language models (LLMs) for solving question-answer (QA) tasks. The state-of-the-art RAG approaches often use the graph data as the external data since they capture the rich semantic information and link relationships between entities. However, existing graph-based RAG approaches cannot accurately identify the relevant information from the graph and also consume large numbers of tokens in the online retrieval process. To address these issues, we introduce a novel graph-based RAG approach, called Attributed Community-based Hierarchical RAG (ArchRAG), by augmenting the question using attributed communities, and also introducing a novel LLM-based hierarchical clustering method. To retrieve the most relevant information from the graph for the question, we build a novel hierarchical index structure for the attributed communities and develop an effective online retrieval method. Experimental results demonstrate that ArchRAG outperforms existing methods in both accuracy and token cost.

While adapting pretrained vision models to downstream dense prediction tasks is widely used, current methods often overlook adaptation efficiency, especially in the context of multi-task learning (MTL). Although parameter-efficient fine-tuning (PEFT) methods can enhance parameter efficiency, broader aspects such as GPU memory and training time efficiency remain underexplored. In this paper, we propose a new paradigm that simultaneously achieves efficiency in Parameters, GPU Memory, and Training Time for Multi-Task Dense Vision Adaptation. Specifically, we propose a dual-branch framework, in which a frozen pretrained backbone serves as the generic main branch, and the proposed Bi-Directional Task Adaptation (BDTA) modules are integrated in parallel to form a task bypass branch that extracts adaptation features required by multiple specific tasks. This adaptation module is lightweight, efficient, and does not require backpropagation through the large pre-trained backbone, thus avoiding resource-intensive gradient computations. Moreover, a Mixture of Task Experts mechanism (MoTE) is further proposed to integrate adaptation features across tasks and scales, thereby obtaining more robust representations tailored for dense prediction tasks. On the PASCAL-Context benchmark, our method achieves over 2× relative performance improvement compared to the best prior multi-task PEFT method, while using only ~30% of the parameters, ~50% of the memory, and ~60% of the training time, demonstrating superior overall adaptation efficiency.

We investigate what structure emerges in 3D Gaussian Splatting (3DGS) solutions from standard multi-view optimization. We term these Rendering-Optimal References (RORs) and analyze their statistical properties, revealing stable patterns-log-normal sca

DNA storage has matured from concept to practical stage, yet its integration with neural compression pipelines remains inefficient. Early DNA encoders applied redundancy-heavy constraint layers atop raw binary data-workable but primitive. Recent neural codecs compress data into learned latent representations with rich statistical structure, yet still convert these latents to DNA via naive binary-to-quaternary transcoding, discarding the entropy model's optimization. This mismatch undermines compression efficiency and complicates the encoding stack. A plug-in module that collapses latent compression and DNA encoding into a single step. SCONE performs quaternary arithmetic coding directly on the latent space in DNA bases. Its Constraint-Aware Adaptive Coding module dynamically steers the entropy encoder's learned probability distribution to enforce biochemical constraints-GC balance and homopolymer suppression-deterministically during encoding, eliminating post-hoc correction. The design preserves full reversibility and exploits the hyperprior model's learned priors without modification. Experiments show SCONE achieves near-perfect constraint satisfaction with negligible computational overhead (< 2% latency), establishing a latent-agnostic interface for end-to-end DNA-compatible learned codecs.

Swizzle is a data access pattern optimization technique by reorganizing the execution order of computational tasks to improve the cache locality in modern NPUs. Existing analysis and optimization techniques lack support for swizzleaware modeling on NPUs and fail to effectively capture cache behavior across diverse swizzle configurations. To this end, we propose TENET-v2, a framework for modeling and optimizing swizzle. We introduce a relation-centric notation to characterize different cache access patterns, thus exploring wider swizzle space. Then, we propose a hybrid performance model for cache analysis. The proposed performance model uses an analytical approach to quantify cache miss behavior under unsaturated cache conditions (non-saturated misses), and employs a simulation method combined with an early exiting mechanism to rapidly model cache behavior under saturated cache conditions (saturated misses). Experimental evaluations demonstrate that TENET-v2 achieves an average absolute error of 1.05 % in read hit rate compared to real-world hardware. Evaluation on a variety of DNNs shows that TENET-v2 outperforms existing tensor program optimizers by up to <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{1. 5} \times$</tex> on A100 GPUs. We also demonstrate NPU cache size optimization based on TENET-v2.

Issue resolution, a complex Software Engineering (SWE) task integral to real-world development, has emerged as a compelling challenge for artificial intelligence. The establishment of benchmarks like SWE-bench revealed this task as profoundly difficul

In this paper, we introduce a novel approach for rigorously verifying safety properties of state machine specifications. Our method leverages an auto-active verifier and centers around the use of action functions annotated with contracts. These contracts facilitate inductive invariant checking, ensuring correctness during system execution. Our approach is further supported by the Why3-do library, which extends the Why3 tool's capabilities to verify concurrent and distributed algorithms using state machines. Two distinctive features of Why3-do are: (i) it supports specification refinement through refinement mappings, enabling hierarchical reasoning about distributed algorithms; and (ii) it can be easily extended to make verifying specific classes of systems more convenient. In particular, the library contains models allowing for message-passing algorithms to be described with programmed handlers , assuming different network semantics. A gallery of examples, all verified with Why3 using SMT solvers as proof tools, is also described in the paper. It contains several auto-actively verified concurrent and distributed algorithms, including the Paxos consensus algorithm.

Abstract Monolithic integration of active and passive photonics is the next step in miniaturization of semiconductor devices, to streamline their manufacturing and in doing so reduce their cost. However, Si’s inefficiency as a light emitter, and the high absorption from quantum wells or dots in III-V materials, requires that the two materials are heterogeneously integrated. This requires either regrowth or additional packaging steps, which add complication in scaling to mass manufacture, as well as high losses at material interfaces. This work proposes a solution whereby a monolithically grown GaAs/AlGaAs platform with both active and passive devices is achieved by using two vertically stacked waveguides, where only one of them contains quantum dots. In the gain section, where the InAs quantum dots emit light centred at 1300 nm, the waveguides are decoupled. By changing the relative environment of the two waveguides, light can be made to evanescently couple into the passive waveguide for low loss passive devices. This change can be brought about either by changing the local etch depth, or by the selective oxidation of Al0.98Ga0.02As underlayers. For the correct geometry, 94% of optical power can be successfully coupled between the active and passive waveguide. This coupling is simulated on FIMMPROP by Photon Design Ltd.