arXiv subjects

Jiaqi Shi

Publications and source records attributed to Jiaqi Shi.

11 recordsLinked to original sources

arXiv2026

CHOW-SLAM: Compact Hybrid Representation with Complementary Overlap Window Optimization for RGB-D SLAM

Simultaneous localization and mapping (SLAM) based on Neural Radiance Fields (NeRF) enables dense, continuous scene reconstruction. However, existing systems operating with limited online resources struggle to simultaneously construct two types of constraints, namely, compact yet discriminative spatial constraints derived from scene representations and persistent temporal constraints derived from historical observations. To address this challenge, we propose CHOW-SLAM, a dense RGB-D SLAM framework that explicitly constructs these complementary spatial and temporal constraints. Spatially, we propose a compact parametric-hash (P-H) hybrid representation that organizes components based on planes and grids across scales in P and H branches. A unified multi-output decoder further aligns the ray termination distributions induced by TSDF and density, preserving geometry and appearance under a compact parameter budget. Temporally, we propose a complementary overlap-window strategy to prevent optimization from being dominated by short-term overlap or weakly related historical observations. Within a fixed budget, the strategy retains recent frames, selects high-overlap local frames, and introduces temporally distributed historical keyframes. Loss-aware keyframe insertion and bundle adjustment scheduling further adapt optimization to tracking quality. In addition, ORB-based tracking and geometric pose estimation are used for pose initialization, followed by neural rendering optimization to improve tracking stability. Extensive evaluations on multiple datasets demonstrate that CHOW-SLAM outperforms state-of-the-art methods in both scene reconstruction quality and camera tracking accuracy. The source code is available at https://github.com/jinjidexiaohuoban/CHOW-SLAM.

cs.CV↗

arXiv2026

Ultra Flash: Scaling Real-Time Streaming Video Generation to High Resolutions

While recent autoregressive video diffusion models achieve remarkable streaming quality, they remain confined to low resolutions (e.g., 480P), leaving efficient, scalable, real-time high-resolution video generation a fundamental open challenge. To bridge this gap, we present Ultra Flash, a cascaded streaming framework capable of real-time high-resolution video generation. Ultra Flash achieves ~30 FPS at 1K resolution and ~18 FPS at 2K resolution on a single GPU through three key contributions: (1) an architecture-preserving T2V-to-TV2V super-resolution training paradigm coupled with an AIGC-oriented data degradation pipeline that effectively preserves the generative capability of the base model, enabling enhanced high-resolution detail when cascaded after mainstream low-resolution generative models; (2) a causal streaming latent upsampler paired with a high-resolution decoder, which enhances spatiotemporal coherence while enabling efficient latent spatial scaling and precise high-resolution decoding with negligible computational overhead; and (3) a cascade high-resolution streaming video generation optimization scheme that first performs hybrid-reward-enhanced sparse causalization and single-step distillation of the super-resolution model, then introduces cascaded streaming self-forcing preference optimization with dynamic cache management, jointly enhancing overall coherence, improving quality, and enabling real-time high-resolution streaming video generation. Extensive experiments demonstrate that Ultra Flash reliably produces ultra-high-resolution streaming video while maintaining state-of-the-art visual quality and superior efficiency. Project Page: https://xin1u.github.io/UltraFlash/

cs.CV↗

arXiv2026

Channel-Level Relation to Attentive Aggregation with Neighborhood-Homogeneity Constraint for Point Cloud Analysis

In 3D point cloud understanding, the core challenge lies in accurately capturing discriminative features within complex neighborhoods, which directly affects the execution precision of downstream tasks such as embodied AI and autonomous driving. Existing methods explore feature correlation discrimination but are limited to point-level spatial distribution or channel responses, enabling only coarse-grained level evaluation. For modern multi-scale point cloud networks, such coarse-grained metrics inevitably incur significant information loss in deeper layers. To address this, we propose PointCRA, a novel network with a channel-level metric-based enhancement mechanism. Our core idea is to introduce temporal trend variation as a new evaluation dimension to avoid the information loss caused by weight dimension collapse in existing spatial and channel attention mechanisms. On this basis, we construct a multi-level calibration framework guided by neighborhood homogeneity for weight calibration, and design a dedicated loss function to enhance channel discriminability.PointCRA leverages intrinsic feature priors to adaptively correct feature aggregation, offering interpretability with low parameter overhead. Our method is transferable, interpretable, and efficient. We validate the proposed method on diverse datasets and benchmark models, and further demonstrate its rationality through extensive analytical experiments. Our PointCRA achieves 77.5\% mIoU on the S3DIS dataset, 90.4\% OA on the ScanObjectNN dataset, and 87.4\% instance mIoU on the ShapeNetPart dataset. The code and pretrained weights are publicly available on GitHub: https://github.com/AGENT9717/PointCRA

cs.CV↗

arXiv2026

From Inheritance to Saturation: Disentangling the Evolution of Visual Redundancy for Architecture-Aware MLLM Inference Acceleration

High-resolution Multimodal Large Language Models (MLLMs) face prohibitive computational costs during inference due to the explosion of visual tokens. Existing acceleration strategies, such as token pruning or layer sparsity, suffer from severe "backbone dependency", performing well on Vicuna or Mistral architectures (e.g., LLaVA) but causing significant performance degradation when transferred to architectures like Qwen. To address this, we leverage truncated matrix entropy to uncover a universal three-stage inference lifecycle, decoupling visual redundancy into universal Intrinsic Visual Redundancy (IVR) and architecture-dependent Secondary Saturation Redundancy (SSR). Guided by this insight, we propose HalfV, a framework that first mitigates IVR via a unified pruning strategy and then adaptively handles SSR based on its specific manifestation. Experiments demonstrate that HalfV achieves superior efficiency-performance trade-offs across diverse backbones. Notably, on Qwen25-VL, it retains 96.8\% performance at a 4.1$\times$ FLOPs speedup, significantly outperforming state-of-the-art baselines. Our code is available at https://github.com/civilizwa/HalfV.

cs.CV↗

arXiv2026

CARE: Multi-Task Pretraining for Latent Continuous Action Representation in Robot Control

Recent advances in Vision-Language-Action (VLA) models have shown promise for robot control, but their dependence on action supervision limits scalability and generalization. To address this challenge, we introduce CARE, a novel framework designed to train VLA models for robotic task execution. Unlike existing methods that depend on action annotations during pretraining, CARE eliminates the need for explicit action labels by leveraging only video-text pairs. These weakly aligned data sources enable the model to learn continuous latent action representations through a newly designed multi-task pretraining objective. During fine-tuning, a small set of labeled data is used to train the action head for control. Experimental results across various simulation tasks demonstrate CARE's superior success rate, semantic interpretability, and ability to avoid shortcut learning. These results underscore CARE's scalability, interpretability, and effectiveness in robotic control with weak supervision.

cs.RO↗

arXiv2025

Enhancing point cloud analysis via neighbor aggregation correction based on cross-stage structure correlation

Point cloud analysis is the cornerstone of many downstream tasks, among which aggregating local structures is the basis for understanding point cloud data. While numerous works aggregate neighbor using three-dimensional relative coordinates, there are irrelevant point interference and feature hierarchy gap problems due to the limitation of local coordinates. Although some works address this limitation by refining spatial description though explicit modeling of cross-stage structure, these enhancement methods based on direct geometric structure encoding have problems of high computational overhead and noise sensitivity. To overcome these problems, we propose the Point Distribution Set Abstraction module (PDSA) that utilizes the correlation in the high-dimensional space to correct the feature distribution during aggregation, which improves the computational efficiency and robustness. PDSA distinguishes the point correlation based on a lightweight cross-stage structural descriptor, and enhances structural homogeneity by reducing the variance of the neighbor feature matrix and increasing classes separability though long-distance modeling. Additionally, we introducing a key point mechanism to optimize the computational overhead. The experimental result on semantic segmentation and classification tasks based on different baselines verify the generalization of the method we proposed, and achieve significant performance improvement with less parameter cost. The corresponding ablation and visualization results demonstrate the effectiveness and rationality of our method. The code and training weight is available at: https://github.com/AGENT9717/PointDistribution

cs.CV↗

arXiv2025

Price and Assortment Optimization under the Multinomial Logit Model with Opaque Products

An opaque product is a product for which only partial information is disclosed to the buyer at the time of purchase. Opaque products are common in sectors such as travel and online retail, where the brand or product color is hidden in the opaque product. Opaque products enable sellers to target customers who prefer a price discount in exchange for being flexible about the product they receive. In this paper, we integrate opaque products and traditional products together into the multinomial logit (MNL) choice model and study the associated price and assortment optimization problems. For the price optimization problem, we show a surprising result that uniform pricing is optimal which implies it has the same optimal pricing solution and value as the traditional MNL model. Although adding an opaque product may enhance revenue given arbitrary traditional product prices, our result shows that this advantage disappears when all prices are optimized jointly. For the assortment problem, we first study the setting where the seller chooses both the offered assortment and the support of the opaque product, and prove that there exists an optimal assortment that is nested by revenue. We then study the case where the opaque support is required to coincide with the offered assortment which is common in practice. Here, we show that the optimal assortment is nested-by-valuation for uniformly priced products. For arbitrary prices, we propose a nested-by-revenue-and-valuation heuristic with a theoretical approximation guarantee that performs extremely well in our numerical study.

math.OC↗

arXiv2024

Vision-Based Deep Reinforcement Learning of UAV Autonomous Navigation Using Privileged Information

The capability of UAVs for efficient autonomous navigation and obstacle avoidance in complex and unknown environments is critical for applications in agricultural irrigation, disaster relief and logistics. In this paper, we propose the DPRL (Distributed Privileged Reinforcement Learning) navigation algorithm, an end-to-end policy designed to address the challenge of high-speed autonomous UAV navigation under partially observable environmental conditions. Our approach combines deep reinforcement learning with privileged learning to overcome the impact of observation data corruption caused by partial observability. We leverage an asymmetric Actor-Critic architecture to provide the agent with privileged information during training, which enhances the model's perceptual capabilities. Additionally, we present a multi-agent exploration strategy across diverse environments to accelerate experience collection, which in turn expedites model convergence. We conducted extensive simulations across various scenarios, benchmarking our DPRL algorithm against the state-of-the-art navigation algorithms. The results consistently demonstrate the superior performance of our algorithm in terms of flight efficiency, robustness and overall success rate.

cs.RO↗

arXiv2024

FunAudioLLM: Voice Understanding and Generation Foundation Models for Natural Interaction Between Humans and LLMs

This report introduces FunAudioLLM, a model family designed to enhance natural voice interactions between humans and large language models (LLMs). At its core are two innovative models: SenseVoice, which handles multilingual speech recognition, emotion recognition, and audio event detection; and CosyVoice, which facilitates natural speech generation with control over multiple languages, timbre, speaking style, and speaker identity. SenseVoice-Small delivers exceptionally low-latency ASR for 5 languages, and SenseVoice-Large supports high-precision ASR for over 50 languages, while CosyVoice excels in multi-lingual voice generation, zero-shot in-context learning, cross-lingual voice cloning, and instruction-following capabilities. The models related to SenseVoice and CosyVoice have been open-sourced on Modelscope and Huggingface, along with the corresponding training, inference, and fine-tuning codes released on GitHub. By integrating these models with LLMs, FunAudioLLM enables applications such as speech-to-speech translation, emotional voice chat, interactive podcasts, and expressive audiobook narration, thereby pushing the boundaries of voice interaction technology. Demos are available at https://fun-audio-llm.github.io, and the code can be accessed at https://github.com/FunAudioLLM.

cs.SD↗

arXiv2023

The Power of Static Pricing for Reusable Resources

We consider the problem of pricing a reusable resource service system. Potential customers arrive according to a Poisson process and purchase the service if their valuation exceeds the current price. If no units are available, customers immediately leave without service. Serving a customer corresponds to using one unit of the reusable resource, where the service time has a general distribution. The objective is to maximize the steady-state revenue rate. This system is equivalent to the classical Erlang loss model with price-sensitive customers, which has applications in vehicle sharing, cloud computing, and spare parts management. With general service times, the optimal pricing policy depends not only on the number of customers currently in the system but also on how long each unavailable unit has been in use. We prove several results that show a simple static policy is universally near-optimal for any service time distribution, arrival rate, and number of units in the system. When there are multiple classes of customers with regular valuation distributions, we prove that static pricing guarantees 78.9% of the revenue of the optimal dynamic policy, achieving the same guarantee known for a single class of customers with exponential service times. When there is one class of customers who have a monotone hazard rate valuation distribution, we prove that a static pricing policy guarantees 97.5% of the revenue from the optimal inventory-based policy. In addition, we prove that the optimal static policy can be easily computed, resulting in the first polynomial-time approximation algorithm for the multi-class problem. Finally, we consider the extensions to irregular valuation distributions and general multi-resource networks.

cs.GT↗

arXiv2020

Multi-View Collaborative Network Embedding

Real-world networks often exist with multiple views, where each view describes one type of interaction among a common set of nodes. For example, on a video-sharing network, while two user nodes are linked if they have common favorite videos in one view, they can also be linked in another view if they share common subscribers. Unlike traditional single-view networks, multiple views maintain different semantics to complement each other. In this paper, we propose MANE, a multi-view network embedding approach to learn low-dimensional representations. Similar to existing studies, MANE hinges on diversity and collaboration - while diversity enables views to maintain their individual semantics, collaboration enables views to work together. However, we also discover a novel form of second-order collaboration that has not been explored previously, and further unify it into our framework to attain superior node representations. Furthermore, as each view often has varying importance w.r.t. different nodes, we propose MANE+, an attention-based extension of MANE to model node-wise view importance. Finally, we conduct comprehensive experiments on three public, real-world multi-view networks, and the results demonstrate that our models consistently outperform state-of-the-art approaches.

cs.LG↗