To achieve low TCP flow completion time (FCT) in data center networks (DCNs), it is critical and challenging to rapidly recover loss without adding extra congestion. Therefore, in this paper, we propose a novel loss recovery approach fast multi-path loss recovery (FUSO) that exploits multi-path diversity in DCN for transport loss recovery. In \name, when a multi-path transport sender suspects loss on one sub-flow, recovery packets are immediately sent over another sub-flow that is not or less lossy and has spare congestion window slots. \name is fast in that it does not need to wait for timeout on the lossy sub-flow, and it is cautious in that it does not violate the congestion control algorithm. Testbed experiments and simulations show that \name decreases the latency-sensitive flows' 99th percentile FCT by up to ~82.3% in a 1-Gb/s testbed, and up to ~87.9% in a 10 Gb/s large-scale simulated network.

Nowadays the privacy issue arising in data mining applications has attracted much attention. In the context of distributed data mining, a major concern of the participant is that its privacy may be disclosed to other participants or a third party. To protect privacy, one can apply a differential privacy approach to perturb the data before sharing them with others, which generally causes a negative effect on the mining result. Thus there is a trade-off between privacy and the mining result. In this paper, we study a distributed classification scenario where a mediator builds a classifier based on the perturbed query results returned by a number of users. We propose a game theoretical approach to analyze how users choose their privacy budgets. Specifically, interactions among users are modeled as a game in satisfaction form. And an algorithm is proposed for users to learn the satisfaction equilibrium (SE) of the game. Experimental results demonstrate that, when the differences among users' expectations are not significant, the proposed learning algorithm can converge to an SE, at which every user achieves a balance between the accuracy of the classifier and the preserved privacy.

The non-consensual exploitation of facial manipulation has emerged as a pressing societal concern. In tandem with the identification of such fake content, recent research endeavors have advocated countering manipulation techniques through proactive interventions, specifically the incorporation of adversarial noise to impede the manipulation in advance. Nevertheless, with insufficient consideration of robustness, we show that current methods falter in providing protection after simple perturbations, e.g., blur. In addition, traditional optimization-based methods face limitations in scalability as they struggle to accommodate the substantial expansion of data volume, a consequence of the time-intensive iterative pipeline. To solve these challenges, we propose a learning-based model, Adversarial Robust Safeguard (ARS), to generate desirable protection noise in a single forward process, concurrently exhibiting a heightened resistance against prevalent perturbations. Specifically, our method involves a two-way protection design, characterized by a basic protection component responsible for generating efficacious noise features, coupled with robust protection for further enhancement. In robust protection, we first fuse image features with spatially duplicated noise embedding, thereby accounting for inherent information redundancy. Subsequently, a combination comprising a differentiable perturbation module and an adversarial network is devised to simulate potential information degradation during the training process. To evaluate it, we conduct experiments on four manipulation methods and compare recent works comprehensively. The results of our method exhibit good visual effects with pronounced robustness against varied perturbations at different levels.

The network intrusion detection system (NIDS) plays an essential role in network security. Although many data-driven approaches from the field of machine learning have been proposed to increase the efficacy of NIDSs, it still suffers from extreme data imbalance and the performance of existing algorithms depends highly on training datasets. To counterpart the class-imbalanced problem in network intrusion detection, it is necessary for models to capture more representative clues within same categories instead of learning from only classification loss. In this paper, we proposed a self-supervised adversarial learning approach for intrusion detection, which utilize instance-level discrimination for better representation learning and employs a adversarial perturbation styled data augmentation to improve the robustness of NIDS on rarely seen attacking types. State-of-the-art result was achieved on multiple frequently-used datasets and experiment conducted on cross-dataset setting demonstrated good generalization ability.

Significant advancements in video question answering (VideoQA) have been made thanks to thriving large image-language pretraining frameworks. Although these image-language models can efficiently represent both video and language branches, they typically employ a goal-free vision perception process and do not interact vision with language well during the answer generation, thus omitting crucial visual cues. In this paper, we are inspired by the human recognition and learning pattern and propose VideoDistill, a framework with language-aware (i.e., goal-driven) behavior in both vision perception and answer generation process. VideoDistill generates answers only from question-related visual embeddings and follows a thinking-observing-answering approach that closely resembles human behavior, distinguishing it from previous research. Specifically, we develop a language-aware gating mechanism to replace the standard cross-attention, avoiding language's direct fusion into visual representations. We incorporate this mechanism into two key components of the entire framework. The first component is a differentiable sparse sampling module, which selects frames containing the necessary dynamics and semantics relevant to the questions. The second component is a vision refinement module that merges existing spatial-temporal attention layers to ensure the extraction of multi-grained visual semantics associated with the questions. We conduct experimental evaluations on various challenging video question-answering benchmarks, and VideoDistill achieves state-of-the-art performance in both general and long-form VideoQA datasets. In Addition, we verify that VideoDistill can effectively alleviate the utilization of language shortcut solutions in the EgoTaskQA dataset.

Teeth localization, segmentation, and labeling in 2D images have great potential in modern dentistry to enhance dental diagnostics, treatment planning, and population-based studies on oral health. However, general instance segmentation frameworks are incompetent due to 1) the subtle differences between some teeth' shapes (e.g., maxillary first premolar and second premolar), 2) the teeth's position and shape variation across subjects, and 3) the presence of abnormalities in the dentition (e.g., caries and edentulism). To address these problems, we propose a ViT-based framework named TeethSEG, which consists of stacked Multi-Scale Aggregation (MSA) blocks and an Anthropic Prior Knowledge (APK) layer. Specifically, to compose the two modules, we design 1) a unique permutation-based upscaler to ensure high efficiency while establishing clear segmentation boundaries with 2) multi-head self/cross-gating layers to emphasize particular semantics meanwhile maintaining the divergence between token embeddings. Besides, we collect 3) the first open-sourced intraoral image dataset IO150K, which comprises over 150k intraoral photos, and all photos are annotated by orthodontists using a human-machine hybrid algorithm. Experiments on IO150K demonstrate that our TeethSEG outperforms the state-of-the-art segmentation models on dental image segmentation.