The NPU-HWC System for the ISCSLP 2024 Inspirational and Convincing Audio Generation Challenge

This paper presents the NPU-HWC system submitted to the ISCSLP 2024 Inspirational and Convincing Audio Generation Challenge 2024 (ICAGC). Our system consists of two modules: a speech generator for Track 1 and a background audio generator for Track 2. In Track 1, we employ Single-Codec to tokenize the speech into discrete tokens and use a language-model-based approach to achieve zero-shot speaking style cloning. The Single-Codec effectively decouples timbre and speaking style at the token level, reducing the acoustic modeling burden on the autoregressive language model. Additionally, we use DSPGAN to upsample 16 kHz mel-spectrograms to high-fidelity 48 kHz waveforms. In Track 2, we propose a background audio generator based on large language models (LLMs). This system produces scene-appropriate accompaniment descriptions, synthesizes background audio with Tango 2, and integrates it with the speech generated by our Track 1 system. Our submission achieves the second place and the first place in Track 1 and Track 2 respectively.

Offline Model-Based Optimization by Learning to Rank

Offline model-based optimization (MBO) aims to identify a design that maximizes a black-box function using only a fixed, pre-collected dataset of designs and their corresponding scores. A common approach in offline MBO is to train a regression-based surrogate model by minimizing mean squared error (MSE) and then find the best design within this surrogate model by different optimizers (e.g., gradient ascent). However, a critical challenge is the risk of out-of-distribution errors, i.e., the surrogate model may typically overestimate the scores and mislead the optimizers into suboptimal regions. Prior works have attempted to address this issue in various ways, such as using regularization techniques and ensemble learning to enhance the robustness of the model, but it still remains. In this paper, we argue that regression models trained with MSE are not well-aligned with the primary goal of offline MBO, which is to select promising designs rather than to predict their scores precisely. Notably, if a surrogate model can maintain the order of candidate designs based on their relative score relationships, it can produce the best designs even without precise predictions. To validate it, we conduct experiments to compare the relationship between the quality of the final designs and MSE, finding that the correlation is really very weak. In contrast, a metric that measures order-maintaining quality shows a significantly stronger correlation. Based on this observation, we propose learning a ranking-based model that leverages learning to rank techniques to prioritize promising designs based on their relative scores. We show that the generalization error on ranking loss can be well bounded. Empirical results across diverse tasks demonstrate the superior performance of our proposed ranking-based models than twenty existing methods.

Spatial Visibility and Temporal Dynamics: Revolutionizing Field of View Prediction in Adaptive Point Cloud Video Streaming

Field-of-View (FoV) adaptive streaming significantly reduces bandwidth requirement of immersive point cloud video (PCV) by only transmitting visible points in a viewer's FoV. The traditional approaches often focus on trajectory-based 6 degree-of-freedom (6DoF) FoV predictions. The predicted FoV is then used to calculate point visibility. Such approaches do not explicitly consider video content's impact on viewer attention, and the conversion from FoV to point visibility is often error-prone and time-consuming. We reformulate the PCV FoV prediction problem from the cell visibility perspective, allowing for precise decision-making regarding the transmission of 3D data at the cell level based on the predicted visibility distribution. We develop a novel spatial visibility and object-aware graph model that leverages the historical 3D visibility data and incorporates spatial perception, neighboring cell correlation, and occlusion information to predict the cell visibility in the future. Our model significantly improves the long-term cell visibility prediction, reducing the prediction MSE loss by up to 50% compared to the state-of-the-art models while maintaining real-time performance (more than 30fps) for point cloud videos with over 1 million points.

Low Latency Point Cloud Rendering with Learned Splatting

Point cloud is a critical 3D representation with many emerging applications. Because of the point sparsity and irregularity, high-quality rendering of point clouds is challenging and often requires complex computations to recover the continuous surface representation. On the other hand, to avoid visual discomfort, the motion-to-photon latency has to be very short, under 10 ms. Existing rendering solutions lack in either quality or speed. To tackle these challenges, we present a framework that unlocks interactive, free-viewing and high-fidelity point cloud rendering. We train a generic neural network to estimate 3D elliptical Gaussians from arbitrary point clouds and use differentiable surface splatting to render smooth texture and surface normal for arbitrary views. Our approach does not require per-scene optimization, and enable real-time rendering of dynamic point cloud. Experimental results demonstrate the proposed solution enjoys superior visual quality and speed, as well as generalizability to different scene content and robustness to compression artifacts. The code is available at https://github.com/huzi96/gaussian-pcloud-render .

GroupCDL: Interpretable Denoising and Compressed Sensing MRI via Learned Group-Sparsity and Circulant Attention

Nonlocal self-similarity within images has become an increasingly popular prior in deep-learning models. Despite their successful image restoration performance, such models remain largely uninterpretable due to their black-box construction. Our previous studies have shown that interpretable construction of a fully convolutional denoiser (CDLNet), with performance on par with state-of-the-art black-box counterparts, is achievable by unrolling a convolutional dictionary learning algorithm. In this manuscript, we seek an interpretable construction of a convolutional network with a nonlocal self-similarity prior that performs on par with black-box nonlocal models. We show that such an architecture can be effectively achieved by upgrading the L1 sparsity prior (soft-thresholding) of CDLNet to an image-adaptive group-sparsity prior (group-thresholding). The proposed learned group-thresholding makes use of nonlocal attention to perform spatially varying soft-thresholding on the latent representation. To enable effective training and inference on large images with global artifacts, we propose a novel circulant-sparse attention. We achieve competitive natural-image denoising performance compared to black-box nonlocal DNNs and transformers. The interpretable construction of our network allows for a straightforward extension to Compressed Sensing MRI (CS-MRI), yielding state-of-the-art performance. Lastly, we show robustness to noise-level mismatches between training and inference for denoising and CS-MRI reconstruction.

Standard compliant video coding using low complexity, switchable neural wrappers

The proliferation of high resolution videos posts great storage and bandwidth pressure on cloud video services, driving the development of next-generation video codecs. Despite great progress made in neural video coding, existing approaches are still far from economical deployment considering the complexity and rate-distortion performance tradeoff. To clear the roadblocks for neural video coding, in this paper we propose a new framework featuring standard compatibility, high performance, and low decoding complexity. We employ a set of jointly optimized neural pre- and post-processors, wrapping a standard video codec, to encode videos at different resolutions. The rate-distorion optimal downsampling ratio is signaled to the decoder at the per-sequence level for each target rate. We design a low complexity neural post-processor architecture that can handle different upsampling ratios. The change of resolution exploits the spatial redundancy in high-resolution videos, while the neural wrapper further achieves rate-distortion performance improvement through end-to-end optimization with a codec proxy. Our light-weight post-processor architecture has a complexity of 516 MACs / pixel, and achieves 9.3% BD-Rate reduction over VVC on the UVG dataset, and 6.4% on AOM CTC Class A1. Our approach has the potential to further advance the performance of the latest video coding standards using neural processing with minimal added complexity.

Structural Constraint Integration in Generative Model for Discovery of Quantum Material Candidates

Billions of organic molecules are known, but only a tiny fraction of the functional inorganic materials have been discovered, a particularly relevant problem to the community searching for new quantum materials. Recent advancements in machine-learning-based generative models, particularly diffusion models, show great promise for generating new, stable materials. However, integrating geometric patterns into materials generation remains a challenge. Here, we introduce Structural Constraint Integration in the GENerative model (SCIGEN). Our approach can modify any trained generative diffusion model by strategic masking of the denoised structure with a diffused constrained structure prior to each diffusion step to steer the generation toward constrained outputs. Furthermore, we mathematically prove that SCIGEN effectively performs conditional sampling from the original distribution, which is crucial for generating stable constrained materials. We generate eight million compounds using Archimedean lattices as prototype constraints, with over 10% surviving a multi-staged stability pre-screening. High-throughput density functional theory (DFT) on 26,000 survived compounds shows that over 50% passed structural optimization at the DFT level. Since the properties of quantum materials are closely related to geometric patterns, our results indicate that SCIGEN provides a general framework for generating quantum materials candidates.

Bits-to-Photon: End-to-End Learned Scalable Point Cloud Compression for Direct Rendering

Point cloud is a promising 3D representation for volumetric streaming in emerging AR/VR applications. Despite recent advances in point cloud compression, decoding and rendering high-quality images from lossy compressed point clouds is still challenging in terms of quality and complexity, making it a major roadblock to achieve real-time 6-Degree-of-Freedom video streaming. In this paper, we address this problem by developing a point cloud compression scheme that generates a bit stream that can be directly decoded to renderable 3D Gaussians. The encoder and decoder are jointly optimized to consider both bit-rates and rendering quality. It significantly improves the rendering quality while substantially reducing decoding and rendering time, compared to existing point cloud compression methods. Furthermore, the proposed scheme generates a scalable bit stream, allowing multiple levels of details at different bit-rate ranges. Our method supports real-time color decoding and rendering of high quality point clouds, thus paving the way for interactive 3D streaming applications with free view points.

PlantTracing: Tracing Arabidopsis Thaliana Apex with CenterTrack

This work applies an encoder-decoder-based machine learning network to detect and track the motion and growth of the flowering stem apex of Arabidopsis Thaliana. Based on the CenterTrack, a machine learning back-end network, we trained a model based on ten time-lapsed labeled videos and tested against three videos.

A Decoupling and Aggregating Framework for Joint Extraction of Entities and Relations

Named Entity Recognition and Relation Extraction are two crucial and challenging subtasks in the field of Information Extraction. Despite the successes achieved by the traditional approaches, fundamental research questions remain open. First, most recent studies use parameter sharing for a single subtask or shared features for both two subtasks, ignoring their semantic differences. Second, information interaction mainly focuses on the two subtasks, leaving the fine-grained informtion interaction among the subtask-specific features of encoding subjects, relations, and objects unexplored. Motivated by the aforementioned limitations, we propose a novel model to jointly extract entities and relations. The main novelties are as follows: (1) We propose to decouple the feature encoding process into three parts, namely encoding subjects, encoding objects, and encoding relations. Thanks to this, we are able to use fine-grained subtask-specific features. (2) We propose novel inter-aggregation and intra-aggregation strategies to enhance the information interaction and construct individual fine-grained subtask-specific features, respectively. The experimental results demonstrate that our model outperforms several previous state-of-the-art models. Extensive additional experiments further confirm the effectiveness of our model.