17 accepted papers
Despite recent advances in training-free text-guided image editing using pretrained rectified flow models, achieving a balance between text adherence and background preservation remains a challenge. Existing methods typically treat image editing as an open-loop trajectory generation task, adopting e
Active perception and manipulation are crucial for robots to interact with complex scenes. Existing methods struggle to unify semantic-driven perception actively with robust, viewpoint-invariant execution accordingly. To this end, we propose SaPaVe, an end-to-end framework that jointly learns these
Vision-Language Models (VLMs) have shown remarkable capabilities in spatial reasoning, yet they remain fundamentally limited to qualitative assessments and lack the computational precision required for real-world robotics. Current approaches fail to leverage metric information from depth sensors and…
Reconstructing atmospheric surface \text CO _2 is crucial for understanding climate dynamics and informing global mitigation strategies. Traditional inversion models achieve precise global \text CO _2 reconstruction but rely heavily on uncertain prior estimates of fluxes and emissions. Inspired by r…
Transparent object grasping remains a persistent challenge in robotics, largely due to the difficulty of acquiring precise 3D information. Conventional optical 3D sensors struggle to capture transparent objects, and machine learning methods are often hindered by their reliance on high-quality datase
With the widespread deployment of generative language models, concerns about safety issues have continuously grown. High-quality fine-tuning data generated from red teaming plays a crucial role in the model’s safety. Recently, automated red teaming approaches have been proposed to create test cases.
We propose a novel sequential Monte Carlo (SMC) method for sampling from unnormalized target distributions based on a reverse denoising diffusion process. While recent diffusion-based samplers simulate the reverse diffusion using approximate score functions, they can suffer from accumulating errors…
Spatial referring is a fundamental capability of embodied robots to interact with the 3D physical world. However, even with the powerful pretrained VLMs, recent approaches are still not qualified to accurately understand the complex 3D scenes and dynamically reason about the instruction-indicated lo…
In a semantic frame resource such as FrameNet, the definition sentence of a frame is essential for humans to understand the meaning of the frame intuitively. Recently, several attempts have been made to induce semantic frames from large corpora, but the cost of creating the definition sentences for…
We explore *implicit opinion extraction* as a new component of aspect-based sentiment analysis (ABSA) systems. Prior work in ABSA has investigated opinion extraction as an important subtask, however, these works only label concise, *explicitly*-stated opinion spans. In this work, we present **Shoes-…
Temporal knowledge graph completion that predicts missing links for incomplete temporal knowledge graphs (TKG) is gaining increasing attention. Most existing works have achieved good results by incorporating time information into static knowledge graph embedding methods. However, they ignore the con…
Temporal knowledge graph (TKG) reasoning that infers missing facts in the future is an essential and challenging task. When predicting a future event, there must be a narrative evolutionary process composed of closely related historical facts to support the event's occurrence, namely fact precursors…
We present TraEDITS, a novel traffic trajectory editing framework for autonomous vehicle testing, which can generate new traffic behaviors by controlling each vehicle interactively to increase the diversity or irregularity of traffic testing data. Given a traffic flow with its original trajectories,
Environmental information is recorded in the multipath propagation, and can be accessed in the form of channel state information (CSI) through commodity WiFi devices. A single CSI reading for event detection is adopted in most existing CSI based indoor monitoring system. However, due to the impact o…
Motion detection as a key component in modern security systems has received an increasing attention recently, but most existing solutions require special installation, calibration, and only have a limited coverage. In this paper, we propose WiDetect, a highly accurate, calibration-free, and low-comp…
Next-generation wireless networks are expected to support exponentially increasing number of users and demands of data, which all rely on the essential media: spectrum. Notice that the 5G networks are featured either by wide bandwidth like mmWave systems, or by large-scale antennas like massive MIMO…
Indoor positioning systems (IPS) based on Wi-Fi signals are gaining popularity recently. IPS based on Received Signal Strength Indicator (RSSI) could only achieve a precision of several meters due to the strong temporal and spatial variation of indoor environment. On the other hand, IPS based on Cha…