61 accepted papers
Direct laser deposition, a specialized form of additive manufacturing, shows good potential in numerous high-value applications such as the repair of aeroengine blades. However, the traditional setup for this technique is bulky and not suited for in-situ repair, requiring the costly disassembly of t…
Dielectric elastomer actuators (DEAs), also recognized as artificial muscle, have been widely developed for the soft locomotion robot. With the complaint skeleton and miniaturized dimension, they are well suited for the narrow space inspection. In this work, we propose a novel low profile (1.1mm) an…
Large Vision-Language Models (LVLMs) often omit or misrepresent critical visual content in generated image captions. Minimizing such information loss will force LVLMs to focus on image details to generate precise descriptions. However, measuring information loss during modality conversion is inheren
Diffusion language models (dLMs) have emerged as a promising paradigm enabling parallel generation, but their learning efficiency lags behind that of autoregressive (AR) language models when trained from scratch. To this end, we study AR-to-dLM conversion, which transforms pretrained AR models into …
As language models become increasingly capable, users expect them to provide not only accurate responses but also behaviors aligned with diverse human preferences across a variety of scenarios. To achieve this, Reinforcement learning (RL) pipelines have begun incorporating multiple rewards, each cap…
Bi-temporal Semantic Change Detection (SCD) is a fundamental analytical framework for capturing state transitions across various domains, with remote sensing being a key application area in this work. However, the prevailing paradigm of comparing two discrete temporal snapshots suffers from inherent
Reinforcement learning with verifiable rewards (RLVR) has become a trending paradigm for training reasoning large language models (LLMs). However, due to the autoregressive decoding nature of LLMs, the rollout process becomes the efficiency bottleneck of RL training, consisting of up to 70\% of the…
Aerial docking of micro unmanned aerial vehicles (UAVs) onto aerial carrier platforms is significantly affected by various disturbances, particularly in the terminal docking phase. The downwash generated by the micro UAV and gust wind can induce considerable disturbances to the carrier, resulting in
End-to-end speech-in, speech-out dialogue systems are emerging as a powerful alternative to traditional ASR–LLM–TTS pipelines but remain prone to hallucinations due to limited factual grounding. While text-based dialogue models have effectively mitigated this issue through tools such as web search A…
Large language models are powerful generalists, yet solving deep and complex problems such as those of the Humanity’s Last Exam (HLE) remains both conceptually challenging and computationally expensive. We show that small orchestrators managing other models and a variety of tools are able to both pu…
While large language models (LLMs) have demonstrated strong performance on factoid question answering, they are still prone to hallucination and untruthful responses, particularly when tasks demand information outside their parametric knowledge. Indeed, truthfulness requires more than accuracy---mod…
Direct laser deposition, a form of additive manufacturing, shows potential in numerous high-value applications such as aeroengine blade repair. However, conventional setups are bulky and unsuitable for in-situ repair, requiring costly engine disassembly. This letter presents a miniaturized high-repe
A multi-zone (typically 8×8) time-of-flight (ToF) sensor offers a low-cost, low-power, and compact solution for range measurement, making it ideal for specialized robotic applications. However, its low resolution limits its usability. Pairing a ToF sensor with a camera enhances depth perception and
Dielectric elastomer actuators (DEAs), also recognised as artificial muscle, have been widely developed for the soft locomotion robot. With the complaint skeleton and miniaturised dimension, they are well suited for the narrow space inspection. In this work, we propose a novel low profile (1.1mm) an
Generative retrieval constitutes an innovative approach in information retrieval, leveraging generative language models(LM) to generate a ranked list of document identifiers (docid) for a given query. It simplifies the retrieval pipeline by replacing the large external index with model parameters. H…
Diffusion models for garment-centric human generation from text or image prompts have garnered emerging attention for their great application potential. However, existing methods often face a dilemma: lightweight approaches, such as adapters, are prone to generate inconsistent textures; while finetu…
Event-based structured light (SL) systems have attracted increasing attention for their potential in high-performance 3D measurement. Despite the inherent HDR capability of event cameras, reflective and absorptive surfaces still cause event cluttering and absence, which produce overexposed and under…
Automatic Video Dubbing (AVD) aims to synthesize speech that matches a character’s speaking style and emotion in silent video clips. However, existing approaches rely on attention mechanisms to learn cross-modal prosodic alignment implicitly, making it challenging to capture subtle prosodic variatio…
We propose Hymba, a family of small language models featuring a hybrid-head parallel architecture that integrates attention mechanisms and state space models (SSMs) within the same layer, offering parallel and complementary processing of the same inputs. In this hybrid-head module, attention heads p…
Hallucinations in large vision-language models (LVLMs) pose significant challenges for real-world applications, as LVLMs may generate responses that appear plausible yet remain inconsistent with the associated visual content. This issue rarely occurs in human cognition. We argue that this discrepanc…
Recent advancements in Large Language Models (LLMs) have spurred interest in numerous applications requiring robust long-range capabilities, essential for processing extensive input contexts and continuously generating extended outputs. As sequence lengths increase, the number of Key-Value (KV) pair…
State space models (SSMs) have emerged as an efficient alternative to Transformer models for language modeling, offering linear computational complexity and constant memory usage as context length increases. However, despite their efficiency in handling long contexts, recent studies have shown that…
Pre-training datasets are typically collected from web content and lack inherent domain divisions. For instance, widely used datasets like Common Crawl do not include explicit domain labels, while manually curating labeled datasets such as The Pile is labor-intensive. Consequently, identifying an op…
Efficient deployment of small language models (SLMs) is essential for numerous real-world applications with stringent latency constraints.While previous work on SLM design has primarily focused on reducing the number of parameters to achieve parameter-optimal SLMs, parameter efficiency does not nece…
Recent advances in reasoning-centric language models have highlighted reinforcement learning (RL) as a promising method for aligning models with verifiable rewards. However, it remains contentious whether RL truly expands a model’s reasoning capabilities or merely amplifies high-reward outputs alrea…
Large Language Models (LLMs) have demonstrated impressive in-context learning (ICL) capabilities from few-shot demonstration exemplars. Recent learning-based demonstration selection methods have proven beneficial to ICL by choosing more useful exemplars. While these methods generally assume they lea…
"In computer vision, it is well-known that a lack of data diversity will impair model performance. In this study, we address the challenges of enhancing the dataset diversity problem in order to benefit various downstream tasks such as object detection and instance segmentation. We propose a simple…
Despite numerous studies of landing on moving ground platforms, landing on midair aerial platforms remains a significant challenge due to aerodynamic interference between Unmanned Aerial Vehicles (UAVs). This letter presents a systematic onboard and real-time trajectory optimization solution for the
This letter presents a 5.2g ultra-lightweight air-ground vehicle capable of passive stable flying and terrestrial cruising. Such a design proposes a passive stability layout, including a single-axis rotor, film dampers, and a stabilizer bar. The advantageous synergy between the rotor and the passive
With their flexible nature, continuum robots offer hyper-redundancy regarding their workspace; their backbone can take many shapes upon a single tip position and orientation. Deciphering which backbone shape to use under certain conditions is crucial to their operation, especially given the rise in
"In this paper, we propose WaterMark Detector (), the first invisible watermark detection method under a black-box and annotation-free setting. is capable of detecting arbitrary watermarks within a given detection dataset using a clean non-watermarked dataset as a reference, without relying on speci…
Large language models (LLMs) are increasingly prevalent in conversational systems due to their advanced understanding and generative capabilities in general contexts. However, their effectiveness in task-oriented dialogues (TOD), which requires not only response generation but also effective dialogu…
Adversarial patch attacks present a significant threat to real-world object detectors due to their practical feasibility. Existing defense methods which rely on attack data or prior knowledge struggle to effectively address a wide range of adversarial patches. In this paper we show two inherent char…
Semantic segmentation models while effective for in-distribution categories face challenges in real-world deployment due to encountering out-of-distribution (OoD) objects. Detecting these OoD objects is crucial for safety-critical applications. Existing methods rely on anomaly scores but choosing a…
We study how down-scaling large language model (LLM) size impacts LLM capabilities. We begin by measuring the effects of weight pruning – a popular technique for reducing model size – on the two abilities of LLMs: (a) recalling facts presented during pre-training and (b) processing information prese…
Continuum robots are flexible mechanisms that enable tasks in cluttered environments where conventional robots cannot operate. However, their few design parameters and task/manufacturing constraints leave small margins for optimization. In this paper, we expand the design space of continuum robots b
Image-based Virtual Try-ON aims to transfer an in-shop garment onto a specific person. Existing methods employ a global warping module to model the anisotropic deformation for different garment parts, which fails to preserve the semantic information of different parts when receiving challenging inpu…
Federated learning (FL) is an effective machine learning paradigm where multiple clients can train models based on heterogeneous data in a decentralized manner without accessing their private data. However, existing FL systems undergo performance deterioration due to feature-level test-time shifts,…
Cold-start problem is one of the most challenging problems for recommender systems. One promising solution to this problem is cross-domain recommendation (CDR) which leverages rich information from an auxiliary source domain to improve the performance of recommender system in the target domain. In p…
As an amphibious system, aerial-terrestrial vehicles have demonstrated remarkable environmental adaptability. In order to expand their synergistic advantages, in this letter, an Aerial-Terrestrial vehicle consisting of a pair of unmanned aerial vehicle (UAV) and unmanned ground vehicle (UGV) is prop
Continuum robots are able of in-situ inspection tasks in cluttered environments and narrow passages, where conventional robots and human operators cannot intervene. However, such intervention often requires the robot to interact with the environment, and the low stiffness and payload of continuum ro
While significant progress has been made in garment transfer, one of the most applicable directions of human-centric image generation, existing works overlook the in-the-wild imagery, presenting severe garment-person misalignment as well as noticeable degradation in fine texture details. This paper,…
Flapping Wing Rotorcraft (FWR) combines flapping and rotating wing motion in one element. Such a hybrid design integrates the high-efficiency characteristics of the rotating wing and the high-lift feature of the flapping wing under low Reynolds number, providing a broader range of simultaneous lift…
Various approaches have been proposed for out-of-distribution (OOD) detection by augmenting models, input examples, training set, and optimization objectives. Deviating from existing work, we have a simple hypothesis that standard off-the-shelf models may already contain sufficient information about…
"Federated Learning aims at training a global model from multiple decentralized devices (i.e. clients) without exchanging their private local data. A key challenge is the handling of non-i.i.d. (independent identically distributed) data across multiple clients that may induce disparities of their lo…
We design deep neural networks (DNNs) and corresponding networks' splittings to distribute DNNs' workload to camera sensors and a centralized aggregator on head-mounted devices to meet system performance targets in inference accuracy and latency under the given hardware resource constraints. To achi…
Spiking Neural Network (SNN) has been recognized as one of the next generation of neural networks. Conventionally, SNN can be converted from a pre-trained ANN by only replacing the ReLU activation to spike activation while keeping the parameters intact. Perhaps surprisingly, in this work we show tha…
Efficient path planning for hyper-redundant continuum and snake-like robots is a challenging task due to limited sensing capabilities, high computational loads, multiple possible solutions, and non-linear models. This letter presents a new approach to snake robot navigation and coiling, with an algo
Due to recent pretrained multilingual representation models, it has become feasible to exploit labeled data from one language to train a cross-lingual model that can then be applied to multiple new languages. In practice, however, we still face the problem of scarce labeled data, leading to subpar r…
User data confidentiality protection is becoming a rising challenge in the present deep learning research. Without access to data, conventional data-driven model compression faces a higher risk of performance degradation. Recently, some works propose to generate images from a specific pretrained mod…
We consider the models of deep multi-task learning with recurrent architectures that exploit regularities across tasks to improve the performance of multiple sequence processing tasks jointly. Most existing architectures are painstakingly customized to learn task relationships for different problems…
We propose Additive Powers-of-Two~(APoT) quantization, an efficient non-uniform quantization scheme for the bell-shaped and long-tailed distribution of weights and activations in neural networks. By constraining all quantization levels as the sum of Powers-of-Two terms, APoT quantization enjoys high…
Sentiment analysis is an area of substantial relevance both in industry and in academia, including for instance in social studies. Although supervised learning algorithms have advanced considerably in recent years, in many settings it remains more practical to apply an unsupervised technique. The la…
To reduce memory footprint and run-time latency, techniques such as neural net-work pruning and binarization have been explored separately. However, it is un-clear how to combine the best of the two worlds to get extremely small and efficient models. In this paper, we, for the first time, define t…
Binary neural networks (BNN) have been studied extensively since they run dramatically faster at lower memory and power consumption than floating-point networks, thanks to the efficiency of bit operations. However, contemporary BNNs whose weights and activations are both single bits suffer from seve…
The standard approach to image instance segmentation is to perform the object detection first, and then segment the object from the detection bounding-box. More recently, deep learning methods like Mask R-CNN perform them jointly. However, little research takes into account the uniqueness of the "hu…
In-situ aeroengine maintenance works (e.g. inspection, repair) are highly beneficial as it can significantly reduce currently accepted maintenance cycle which is extensive and costly due to the need to remove engines from the wing of an aircraft. However, feeding in/out via inspection ports and perf…
Precise actuation of continuum surfaces in combination with continuum robotic arms that undergo large deformation is of high interest in soft robotics but of limited model-based study to date. This work develops this area towards enabling the robust design and control of large deformation continuum
Precise actuation of continuum surfaces in combination with continuum robotic arms that undergo large deformation is of high interest in soft robotics but of limited model-based study to date. This work develops this area towards enabling the robust design and control of large deformation continuum…
Automation of inspection and repair tasks on complex installations is gaining attention from industries with high-value assets such as aerospace, nuclear and marine. This paper reports on a five degrees of freedom robotic system capable of performing accurate and repeatable repair procedures through…
How to develop slim and accurate deep neural networks has become crucial for real- world applications, especially for those employed in embedded systems. Though previous work along this research line has shown some promising results, most existing methods either fail to significantly compress a well…