搜索结果：streamed

We investigate incentives for reducing the carbon emissions of video streaming that depend on the energy consumption of segments in the end-to-end video delivery path, the carbon intensity, and the user type, i.e., quality-sensitive and green or environmentally conscious users. The incentives can be offered through a practical 2-tier subscription model with a discount and carbon rewards, which gives providers the flexibility to reduce the quality for up to a maximum percentage of videos within a time period, such as one month. The key features of our approach are i) it is preferable to offer subscriptions where the reduced-quality tier is set one resolution level below the resolution required for maximum user satisfaction; ii) when a video is streamed from a local data center, the maximum percentage of videos streamed at a lower quality depends solely on the carbon intensity and the average intensity cap, whereas the incentives also depend on the users' level of environmental consciousness; iii) when a video can be streamed from a local or a remote data center with different carbon intensities, the maximum percentage of videos streamed at lower quality and the incentives depend on

Large language models increasingly stream long, reasoning-intensive responses in real time, making when to moderate as critical as whether to moderate. Existing guardrails fall into two unsatisfactory extremes: response-level methods delay intervention until the full output is generated, whereas token-level methods act on incomplete semantics, often producing unstable decisions and excessive guard invocations. To address this challenge, we propose SentGuard, a sentence-level streaming guardrail that operates in parallel with generation. A lightweight waiting buffer groups streamed tokens into sentence chunks and releases only verified chunks to the user, introducing a small offset that enables SentGuard to assess the current prefix while the target LLM decodes subsequent content. To support this, we construct StreamSafe, a benchmark with structured per-sentence annotations across 8 harm categories, capturing the evolution of safety risks across both reasoning and response segments. We further train SentGuard with a coarse-to-fine objective to detect unsafe intent as soon as it emerges at sentence boundaries. Experiments on 5 safety benchmarks show that SentGuard outperforms existin

We formalize action emergence as a target capability for end-to-end autonomous driving: the ability to generate physically feasible, semantically appropriate, and safety-compliant actions in arbitrary, long-tail traffic scenes through scene-conditioned reasoning rather than retrieval or interpolation of learned scene-action mappings. We show that previous paradigms cannot deliver action emergence: autoregressive trajectory decoders collapse the inherently multimodal future into a single averaged output, while diffusion and flow-matching generators express multimodality but are not steerable by reasoned intent. We propose Streaming Intent as a concrete way to approach action emergence: a mechanism that makes driving intent (i) semantically streamed through a continuous chain-of-thought that causally derives the intent from scene understanding, and (ii) temporally streamed across clips so that intent commitments remain coherent along the driving horizon. We realize Streaming Intent in a VLA model we call SI (Streaming Intent). SI autoregressively decodes a four-step chain-of-thought and emits an intent token; the decoded intent then drives classifier-free guidance (CFG) on a flow-mat

Cloud rendering is widely used in gaming and XR to overcome limited client-side GPU resources and to support heterogeneous devices. Existing systems typically deliver the rendered scene as a 2D video stream, which tightly couples the transmitted content to the server-rendered viewpoint and limits latency compensation to image-space reprojection or warping. In this paper, we investigate an alternative approach based on streaming a live 3D Gaussian Splatting (3DGS) scene representation instead of only rendered video. We present a Unity-based prototype in which a server constructs and continuously optimizes a 3DGS model from real-time rendered reference views, while streaming the evolving representation to remote clients using full model snapshots and incremental updates supporting relighting and rigid object dynamics. The clients reconstruct the streamed Gaussian model locally and render their current viewpoint from the received representation. This approach aims to improve viewpoint flexibility for latency compensation and to better amortize server-side scene modeling across multiple users than per-user rendering and video streaming. We describe the system design, evaluate it, and c

While video streaming understanding has made significant strides, real-world applications, such as live sports broadcasting, autonomous driving, and multi-screen collaboration, inherently demand continuous, multi-stream interactions. However, existing benchmarks are confined to single-stream paradigms, leaving a critical gap in evaluating online, cross-stream reasoning. To bridge this, we introduce X-Stream, the first benchmark dedicated to multi-stream streaming understanding. Comprising 4,220 rigorously curated QA pairs across 932 videos, X-Stream evaluates 11 subtasks across multi-window, multi-view, and multi-device scenarios. Crucially, our dataset is constructed using a novel dual-verification pipeline that prevents over-reliance on a single stream. Furthermore, we pioneer the conceptualization of multi-modal large language models (MLLMs) as naive multiplexers, systematically evaluating their performance through the lens of Signal Multiplexing Theory. Our extensive online inference experiments reveal a stark reality: state-of-the-art MLLMs struggle significantly with concurrent streams, achieving only about 50% score and exhibiting poor proactive ability. Ultimately, X-Stream

The increasing use of Internet-of-Things (IoT) sensors in moving objects has resulted in vast amounts of spatiotemporal streaming data. To analyze this data in situ, real-time spatiotemporal processing is needed. However, current stream processing systems designed for IoT environments often lack spatiotemporal processing capabilities, and existing spatiotemporal libraries primarily focus on analyzing historical data. This gap makes performing real-time spatiotemporal analytics challenging. In this demonstration, we present NebulaMEOS, which combines MEOS (Mobility Engine Open Source), a spatiotemporal processing library, with NebulaStream, a scalable data management system for IoT applications. By integrating MEOS into NebulaStream, NebulaMEOS utilizes spatiotemporal functionalities to process and analyze streaming data in real-time. We demonstrate NebulaMEOS by querying data streamed from edge devices on trains by the Société Nationale des Chemins de fer Belges (SNCB). Visitors can experience demonstrations of geofencing and geospatial complex event processing, visualizing real-time train operations and environmental impacts.

While ML model training and inference are both GPU-intensive, CPU-based data processing is often the bottleneck. Distributed data processing systems based on the batch or stream processing models assume homogeneous resource requirements. They excel at CPU-based computation but either under-utilize heterogeneous resources or impose high overheads on failure and reconfiguration. We introduce the streaming batch model, a hybrid of batch and streaming that enables efficient and fault-tolerant heterogeneous execution. The key idea is to use partitions as the unit of execution to achieve elasticity, but to allow partitions to be dynamically created and streamed between heterogeneous operators for memory-efficient pipelining. We present Ray Data, a streaming batch system that improves throughput on heterogeneous batch inference pipelines by 2.5-12$\times$ compared to traditional batch and stream processing systems. By leveraging heterogeneous clusters, Ray Data improves training throughput for multimodal models such as Stable Diffusion by 31% compared to single-node ML data loaders.

The increase in video streaming has presented a challenge of handling stream request effectively, especially over networks that are variable. This paper describes a new adaptive video streaming architecture capable of changing the video quality and buffer size depending on the data and latency of streamed video. For video streaming VLC media player was used where network performance data were obtained through Python scripts with very accurate data rate and latency measurement. The collected data is analyzed using Gemini AI, containing characteristics of the machine learning algorithm that recognizes the best resolution of videos and the buffer sizes. Through the features of real-time monitoring and artificial intelligence decision making, the proposed framework improves the user experience by reducing the occurrence of buffering events while at the same time increasing the video quality. Our findings therefore confirm that the proposed solution based on artificial intelligence increases video quality and flexibility. This study advances knowledge of adaptive streaming and offers an argument about how intelligent datadriven approaches and AI may be useful tools for enhancing the del

In this paper, we introduce a large model-empowered streaming semantic communication system for speech transmission across various languages, named LSSC-ST. Specifically, we devise an edge-device collaborative semantic communication architecture by offloading the intricate semantic extraction and channel coding modules to edge servers, thereby reducing the computational burden on local devices. To support multilingual speech transmission, pre-trained large speech models are utilized to learn unified semantic features from speech in different languages, breaking the constraint of a single input language and enhancing the practicality of the LSSC-ST. Moreover, the input speech is sequentially streamed into the developed system as short speech segments, which enables low transmission latency without degrading the quality of the produced speech. A novel dynamic speech segmentation algorithm is proposed to further reduce the transmission latency by adaptively adjusting the duration of speech segments. According to simulation results, the LSSC-ST provides more accurate speech transmission and achieves a streaming manner with lower latency compared to the existing non-streaming semantic c

There has been increasing interest in unifying streaming and non-streaming automatic speech recognition (ASR) models to reduce development, training, and deployment costs. We present a unified framework that trains a single end-to-end ASR model for both streaming and non-streaming applications, leveraging future context information. We propose to use dynamic right-context through the chunked attention masking in the training of zipformer-based ASR models. We demonstrate that using right-context is more effective in zipformer models compared to other conformer models due to its multi-scale nature. We analyze the effect of varying the number of right-context frames on accuracy and latency of the streaming ASR models. We use Librispeech and large in-house conversational datasets to train different versions of streaming and non-streaming models and evaluate them in a production grade server-client setup across diverse testsets of different domains. The proposed strategy reduces word error by relative 7.9\% with a small degradation in user-perceived latency. By adding more right-context frames, we are able to achieve streaming performance close to that of non-streaming models. Our appro

We introduce Delayed Streams Modeling (DSM), a flexible formulation for streaming, multimodal sequence-to-sequence learning. Sequence-to-sequence generation is often cast in an offline manner, where the model consumes the complete input sequence before generating the first output timestep. Alternatively, streaming sequence-to-sequence rely on learning a policy for choosing when to advance on the input stream, or write to the output stream. DSM instead models already time-aligned streams with a decoder-only language model. By moving the alignment to a pre-processing step,and introducing appropriate delays between streams, DSM provides streaming inference of arbitrary output sequences, from any input combination, making it applicable to many sequence-to-sequence problems. In particular, given text and audio streams, automatic speech recognition (ASR) corresponds to the text stream being delayed, while the opposite gives a text-to-speech (TTS) model. We perform extensive experiments for these two major sequence-to-sequence tasks, showing that DSM provides state-of-the-art performance and latency while supporting arbitrary long sequences, being even competitive with offline baselines.

Disaggregation is an ongoing trend to increase flexibility in datacenters. With interconnect technologies like CXL, pools of CPUs, accelerators, and memory can be connected via a datacenter fabric. Applications can then pick from those pools the resources necessary for their specific workload. However, this vision becomes less clear when we consider data movement. Workloads often require data to be streamed through chains of multiple devices, but typically, these data streams physically do not directly flow device-to-device, but are staged in memory by a CPU hosting device protocol logic. We show that augmenting devices with a disaggregation-native and device-independent data streaming facility can improve processing latencies by enabling data flows directly between arbitrary devices.