Video understanding plays a fundamental role for content moderation on short video platforms, enabling the detection of inappropriate content. While classification remains the dominant approach for content moderation, it often struggles in scenarios requiring rapid and cost-efficient responses, such as trend adaptation and urgent escalations. To address this issue, we introduce an Embedding-Based Retrieval (EBR) method designed to complement traditional classification approaches. We first leverage a Supervised Contrastive Learning (SCL) framework to train a suite of foundation embedding models, including both single-modal and multi-modal architectures. Our models demonstrate superior performance over established contrastive learning methods such as CLIP and MoCo. Building on these embedding models, we design and implement the embedding-based retrieval system that integrates embedding generation and video retrieval to enable efficient and effective trend handling. Comprehensive offline experiments on 25 diverse emerging trends show that EBR improves ROC-AUC from 0.85 to 0.99 and PR-AUC from 0.35 to 0.95. Further online experiments reveal that EBR increases action rates by 10.32% and reduces operational costs by over 80%, while also enhancing interpretability and flexibility compared to classification-based solutions.

Dense features, customized for different business scenarios, are essential in short video classification. However, their complexity, specific adaptation requirements, and high computational costs make them resource-intensive and less accessible during online inference. Consequently, these dense features are categorized as 'Privileged Dense Features'.Meanwhile, end-to-end multi-modal models have shown promising results in numerous computer vision tasks. In industrial applications, prioritizing end-to-end multi-modal features, can enhance efficiency but often leads to the loss of valuable information from historical privileged dense features.To integrate both features while maintaining efficiency and manageable resource costs, we present Confidence-aware Privileged Feature Distillation (CPFD), which empowers features of an end-to-end multi-modal model by adaptively distilling privileged features during training.Unlike existing privileged feature distillation (PFD) methods, which apply uniform weights to all instances during distillation, potentially causing unstable performance across different business scenarios and a notable performance gap between teacher model (Dense Feature enhanced multimodal-model DF-X-VLM) and student model (multimodal-model only X-VLM), our CPFD leverages confidence scores derived from the teacher model to adaptively mitigate the performance variance with the student model. We conducted extensive offline experiments on five diverse tasks demonstrating that CPFD improves the video classification F1 score by 6.76% compared with end-to-end multimodal-model (X-VLM) and by 2.31% with vanilla PFD on-average. And it reduces the performance gap by 84.6% and achieves results comparable to teacher model DF-X-VLM. The effectiveness of CPFD is further substantiated by online experiments, and our framework has been deployed in production systems for over a dozen models.

During the one minute it takes to read this abstract, two billion smartphones worldwide will perform billions of Wifi channel scans recording the signal strength of nearby Wifi Access Points (APs). Yet despite this ongoing planetary-scale wireless network measurement, few systematic efforts are made today to recover this potentially valuable data. In this paper we ask the question: “Are the smartphone channel scans useful in monitoring enterprise Wifi networks?” More specifically, can these client-side measurements provide new insights compared to the AP-side measurements that enterprise Wifi networks already perform? Beginning with two Wifi scan datasets collected on two large scale smartphone testbeds, we conduct case studies that show how smartphone channel scans can be used to (1) improve AP spectrum management, and (2) predict the impact of AP failure or overload. In each case, a walk on the client side yields valuable insights for network operators that are otherwise impossible to gain from AP-side measurements, and together our results demonstrate the value of smartphone channel scans.

The hundreds of millions of deployed smartphones provide an unprecedented opportunity to collect data to monitor, debug, and continuously adapt wireless networks to improve performance. In contrast with previous mobile devices, such as laptops, smartphones are always on but mostly idle, making them available to perform measurements that help other nearby active devices make better use of available network resources. We present the design of PocketSniffer, a system delivering wireless measurements from smartphones both to network administrators for monitoring and debugging purposes and to algorithms performing realtime network adaptation. By collecting data from smartphones, PocketSniffer supports novel adaptation algorithms designed around common deployment scenarios involving both cooperative and self-interested clients and networks. We present preliminary results from a prototype and discuss challenges to realizing this vision.

The rising core count per processor is pushing chip complexity to a level that hardware-based cache coherency protocols become too hard and costly to scale someday. We need new designs of many-core hardware and software other than traditional technologies to keep up with the ever-increasing scalability demands. A cluster-on-chip architecture, as exemplified by the Intel Single-chip Cloud Computer (SCC), promotes a software-oriented approach instead of hardware support to implementing shared memory coherence. This paper presents a shared virtual memory (SVM) system, dubbed Rhymes, tailored to new processor kinds of non-coherent and hybrid memory architectures. Rhymes features a two-way cache coherence protocol to enforce release consistency for pages allocated in shared physical memory (SPM) and scope consistency for pages in percore private memory. It also supports page remapping on a percore basis to boost data locality. We implement and test Rhymes on the SCC port of the Barrelfish OS. Experimental results show that our SVM outperforms the pure SPM approach used by Intel's software managed coherence (SMC) library by up to 12 times through improved cache utilization for applications with strong data reuse patterns.