Over the past two decades, the surge in video streaming applications has been fueled by the increasing accessibility of the internet and the growing demand for network video. As users with varying internet speeds and devices seek high-quality video, transcoding becomes essential for service providers. In this paper, we introduce a parametric rate-distortion (R-D) transcoding model. Our model excels at predicting transcoding distortion at various rates without the need for encoding the video. This model serves as a versatile tool that can be used to achieve visual quality improvement (in terms of PSNR) via trans-sizing. Moreover, we use our model to identify visually lossless and near-zero-slope bitrate ranges for an ingest video. Having this information allows us to adjust the transcoding target bitrate while introducing visually negligible quality degradations. By utilizing our model in this manner, quality improvements up to 2 dB and bitrate savings of up to 46% of the original target bitrate are possible. Experimental results demonstrate the efficacy of our model in video transcoding rate distortion prediction.
Dataset distillation aims to compress information from a large-scale original dataset to a new compact dataset while striving to preserve the utmost degree of the original data informational essence. Previous studies have predominantly concentrated on aligning the intermediate statistics between the original and distilled data, such as weight trajectory, features, gradient, BatchNorm, etc. In this work, we consider addressing this task through the new lens of model informativeness in the compression stage on the original dataset pretraining. We observe that with the prior state-of-the-art SRe$^2$L, as model sizes increase, it becomes increasingly challenging for supervised pretrained models to recover learned information during data synthesis, as the channel-wise mean and variance inside the model are flatting and less informative. We further notice that larger variances in BN statistics from self-supervised models enable larger loss signals to update the recovered data by gradients, enjoying more informativeness during synthesis. Building on this observation, we introduce SC-DD, a simple yet effective Self-supervised Compression framework for Dataset Distillation that facilitates diverse information compression and recovery compared to traditional supervised learning schemes, further reaps the potential of large pretrained models with enhanced capabilities. Extensive experiments are conducted on CIFAR-100, Tiny-ImageNet and ImageNet-1K datasets to demonstrate the superiority of our proposed approach. The proposed SC-DD outperforms all previous state-of-the-art supervised dataset distillation methods when employing larger models, such as SRe$^2$L, MTT, TESLA, DC, CAFE, etc., by large margins under the same recovery and post-training budgets. Code is available at https://github.com/VILA-Lab/SRe2L/tree/main/SCDD/.
This study introduces a novel BERT-LSH model that incorporates Locality Sensitive Hashing (LSH) to approximate the attention mechanism in the BERT architecture. We examine the computational efficiency and performance of this model compared to a standard baseline BERT model. Our findings reveal that BERT-LSH significantly reduces computational demand for the self-attention layer while unexpectedly outperforming the baseline model in pretraining and fine-tuning tasks. These results suggest that the LSH-based attention mechanism not only offers computational advantages but also may enhance the model's ability to generalize from its training data. For more information, visit our GitHub repository: https://github.com/leo4life2/algoml-final
In this paper, we have expanded the current status of semantic communication limited to processing one task to a more general system that can handle multiple tasks concurrently. In pursuit of this, we first introduced our definition of the "semantic source", enabling the interpretation of multiple semantics based on a single observation. A semantic encoder design is then introduced, featuring the division of the encoder into a common unit and multiple specific units enabling cooperative multi-task processing. Simulation results demonstrate the effectiveness of the proposed semantic source and the system design. Our approach employs information maximization (infomax) and end-to-end design principles.
Over the past few years, we have witnessed remarkable advancements in Code Pre-trained Models (CodePTMs). These models achieved excellent representation capabilities by designing structure-based pre-training tasks for code. However, how to enhance the absorption of structural knowledge when fine-tuning CodePTMs still remains a significant challenge. To fill this gap, in this paper, we present Structure-aware Fine-tuning (SAT), a novel structure-enhanced and plug-and-play fine-tuning method for CodePTMs. We first propose a structure loss to quantify the difference between the information learned by CodePTMs and the knowledge extracted from code structure. Specifically, we use the attention scores extracted from Transformer layer as the learned structural information, and the shortest path length between leaves in abstract syntax trees as the structural knowledge. Subsequently, multi-task learning is introduced to improve the performance of fine-tuning. Experiments conducted on four pre-trained models and two generation tasks demonstrate the effectiveness of our proposed method as a plug-and-play solution. Furthermore, we observed that SAT can benefit CodePTMs more with limited training data.
Despite previous DETR-like methods having performed successfully in generic object detection, tiny object detection is still a challenging task for them since the positional information of object queries is not customized for detecting tiny objects, whose scale is extraordinarily smaller than general objects. Also, DETR-like methods using a fixed number of queries make them unsuitable for aerial datasets, which only contain tiny objects, and the numbers of instances are imbalanced between different images. Thus, we present a simple yet effective model, named DQ-DETR, which consists of three different components: categorical counting module, counting-guided feature enhancement, and dynamic query selection to solve the above-mentioned problems. DQ-DETR uses the prediction and density maps from the categorical counting module to dynamically adjust the number of object queries and improve the positional information of queries. Our model DQ-DETR outperforms previous CNN-based and DETR-like methods, achieving state-of-the-art mAP 30.2% on the AI-TOD-V2 dataset, which mostly consists of tiny objects.
Multimodal video sentiment analysis aims to integrate multiple modal information to analyze the opinions and attitudes of speakers. Most previous work focuses on exploring the semantic interactions of intra- and inter-modality. However, these works ignore the reliability of multimodality, i.e., modalities tend to contain noise, semantic ambiguity, missing modalities, etc. In addition, previous multimodal approaches treat different modalities equally, largely ignoring their different contributions. Furthermore, existing multimodal sentiment analysis methods directly regress sentiment scores without considering ordinal relationships within sentiment categories, with limited performance. To address the aforementioned problems, we propose a trustworthy multimodal sentiment ordinal network (TMSON) to improve performance in sentiment analysis. Specifically, we first devise a unimodal feature extractor for each modality to obtain modality-specific features. Then, an uncertainty distribution estimation network is customized, which estimates the unimodal uncertainty distributions. Next, Bayesian fusion is performed on the learned unimodal distributions to obtain multimodal distributions for sentiment prediction. Finally, an ordinal-aware sentiment space is constructed, where ordinal regression is used to constrain the multimodal distributions. Our proposed TMSON outperforms baselines on multimodal sentiment analysis tasks, and empirical results demonstrate that TMSON is capable of reducing uncertainty to obtain more robust predictions.
The primary focus of Neural Representation for Videos (NeRV) is to effectively model its spatiotemporal consistency. However, current NeRV systems often face a significant issue of spatial inconsistency, leading to decreased perceptual quality. To address this issue, we introduce the Pyramidal Neural Representation for Videos (PNeRV), which is built on a multi-scale information connection and comprises a lightweight rescaling operator, Kronecker Fully-connected layer (KFc), and a Benign Selective Memory (BSM) mechanism. The KFc, inspired by the tensor decomposition of the vanilla Fully-connected layer, facilitates low-cost rescaling and global correlation modeling. BSM merges high-level features with granular ones adaptively. Furthermore, we provide an analysis based on the Universal Approximation Theory of the NeRV system and validate the effectiveness of the proposed PNeRV.We conducted comprehensive experiments to demonstrate that PNeRV surpasses the performance of contemporary NeRV models, achieving the best results in video regression on UVG and DAVIS under various metrics (PSNR, SSIM, LPIPS, and FVD). Compared to vanilla NeRV, PNeRV achieves a +4.49 dB gain in PSNR and a 231% increase in FVD on UVG, along with a +3.28 dB PSNR and 634% FVD increase on DAVIS.
The proliferation of Large Language Models (LLMs) highlights the critical importance of conducting thorough evaluations to discern their comparative advantages, limitations, and optimal use cases. Particularly important is assessing their capacity to accurately retrieve information included in a given prompt. A model's ability to do this significantly influences how effectively it can utilize contextual details, thus impacting its practical efficacy and dependability in real-world applications. Our research analyzes the in-context recall performance of various LLMs using the needle-in-a-haystack method. In this approach, a factoid (the "needle") is embedded within a block of filler text (the "haystack"), which the model is asked to retrieve. We assess the recall performance of each model across various haystack lengths and with varying needle placements to identify performance patterns. This study demonstrates that an LLM's recall capability is not only contingent upon the prompt's content but also may be compromised by biases in its training data. Conversely, adjustments to model architecture, training strategy, or fine-tuning can improve performance. Our analysis provides insight into LLM behavior, offering direction for the development of more effective applications of LLMs.
The proliferation of unreliable news domains on the internet has had wide-reaching negative impacts on society. We introduce and evaluate interventions aimed at reducing traffic to unreliable news domains from search engines while maintaining traffic to reliable domains. We build these interventions on the principles of fairness (penalize sites for what is in their control), generality (label/fact-check agnostic), targeted (increase the cost of adversarial behavior), and scalability (works at webscale). We refine our methods on small-scale webdata as a testbed and then generalize the interventions to a large-scale webgraph containing 93.9M domains and 1.6B edges. We demonstrate that our methods penalize unreliable domains far more than reliable domains in both settings and we explore multiple avenues to mitigate unintended effects on both the small-scale and large-scale webgraph experiments. These results indicate the potential of our approach to reduce the spread of misinformation and foster a more reliable online information ecosystem. This research contributes to the development of targeted strategies to enhance the trustworthiness and quality of search engine results, ultimately benefiting users and the broader digital community.