Quadratic Unconstrained Binary Optimization (QUBO) sits at the heart of many industries and academic fields such as logistics, supply chain, finance, pharmaceutical science, chemistry, IT, and energy sectors, among others. These problems typically involve optimizing a large number of binary variables, which makes finding exact solutions exponentially more difficult. Consequently, most QUBO problems are classified as NP-hard. To address this challenge, we developed a powerful feedforward neural network (FNN) optimizer for arbitrary QUBO problems. In this work, we demonstrate that the FNN optimizer can provide high-quality approximate solutions for large problems, including dense 80-variable weighted MaxCut and random QUBOs, achieving an average accuracy of over 99% in less than 1.1 seconds on an 8-core CPU. Additionally, the FNN optimizer outperformed the Gurobi optimizer by 72% on 200-variable random QUBO problems within a 100-second computation time limit, exhibiting strong potential for real-time optimization tasks. Building on this model, we explored the novel approach of integrating FNNs with a quantum annealer-based activation function to create a quantum-classical encoder-decoder (QCED) optimizer, aiming to further enhance the performance of FNNs in QUBO optimization.
In recent years, much interdisciplinary research has been conducted exploring potential use cases of neuroscience to advance the field of information retrieval. Initial research concentrated on the use of fMRI data, but fMRI was deemed to be not suitable for real-world applications, and soon, research shifted towards using EEG data. In this paper, we try to improve the original performance of a first attempt at generating text using EEG by focusing on the less explored area of optimising neural network performance. We test a set of different activation functions and compare their performance. Our results show that introducing a higher degree polynomial activation function can enhance model performance without changing the model architecture. We also show that the learnable 3rd-degree activation function performs better on the 1-gram evaluation compared to a 3rd-degree non-learnable function. However, when evaluating the model on 2-grams and above, the polynomial function lacks in performance, whilst the leaky ReLU activation function outperforms the baseline.
This paper presents an advanced end-to-end singing voice synthesis (SVS) system based on the source-filter mechanism that directly translates lyrical and melodic cues into expressive and high-fidelity human-like singing. Similarly to VISinger 2, the proposed system also utilizes training paradigms evolved from VITS and incorporates elements like the fundamental pitch (F0) predictor and waveform generation decoder. To address the issue that the coupling of mel-spectrogram features with F0 information may introduce errors during F0 prediction, we consider two strategies. Firstly, we leverage mel-cepstrum (mcep) features to decouple the intertwined mel-spectrogram and F0 characteristics. Secondly, inspired by the neural source-filter models, we introduce source excitation signals as the representation of F0 in the SVS system, aiming to capture pitch nuances more accurately. Meanwhile, differentiable mcep and F0 losses are employed as the waveform decoder supervision to fortify the prediction accuracy of speech envelope and pitch in the generated speech. Experiments on the Opencpop dataset demonstrate efficacy of the proposed model in synthesis quality and intonation accuracy.
Scaling laws offer valuable insights into the design of time series foundation models (TSFMs). However, previous research has largely focused on the scaling laws of TSFMs for in-distribution (ID) data, leaving their out-of-distribution (OOD) scaling behavior and the influence of model architectures less explored. In this work, we examine two common TSFM architectures, encoder-only and decoder-only Transformers, and investigate their scaling behavior on both ID and OOD data. These models are trained and evaluated across varying parameter counts, compute budgets, and dataset sizes. Our experiments reveal that the log-likelihood loss of TSFMs exhibits similar scaling behavior in both OOD and ID settings. We further compare the scaling properties across different architectures, incorporating two state-of-the-art TSFMs as case studies, showing that model architecture plays a significant role in scaling. The encoder-only Transformers demonstrate better scalability than the decoder-only Transformers, while the architectural enhancements in the two advanced TSFMs primarily improve ID performance but reduce OOD scalability. While scaling up TSFMs is expected to drive performance breakthroughs, the lack of a comprehensive understanding of TSFM scaling laws has hindered the development of a robust framework to guide model scaling. We fill this gap in this work by synthesizing our findings and providing practical guidelines for designing and scaling larger TSFMs with enhanced model capabilities.
The development of efficient surrogates for partial differential equations (PDEs) is a critical step towards scalable modeling of complex, multiscale systems-of-systems. Convolutional neural networks (CNNs) have gained popularity as the basis for such surrogate models due to their success in capturing high-dimensional input-output mappings and the negligible cost of a forward pass. However, the high cost of generating training data -- typically via classical numerical solvers -- raises the question of whether these models are worth pursuing over more straightforward alternatives with well-established theoretical foundations, such as Monte Carlo methods. To reduce the cost of data generation, we propose training a CNN surrogate model on a mixture of numerical solutions to both the $d$-dimensional problem and its ($d-1$)-dimensional approximation, taking advantage of the efficiency savings guaranteed by the curse of dimensionality. We demonstrate our approach on a multiphase flow test problem, using transfer learning to train a dense fully-convolutional encoder-decoder CNN on the two classes of data. Numerical results from a sample uncertainty quantification task demonstrate that our surrogate model outperforms Monte Carlo with several times the data generation budget.
Current test- or compression-time adaptation image compression (TTA-IC) approaches, which leverage both latent and decoder refinements as a two-step adaptation scheme, have potentially enhanced the rate-distortion (R-D) performance of learned image compression models on cross-domain compression tasks, \textit{e.g.,} from natural to screen content images. However, compared with the emergence of various decoder refinement variants, the latent refinement, as an inseparable ingredient, is barely tailored to cross-domain scenarios. To this end, we aim to develop an advanced latent refinement method by extending the effective hybrid latent refinement (HLR) method, which is designed for \textit{in-domain} inference improvement but shows noticeable degradation of the rate cost in \textit{cross-domain} tasks. Specifically, we first provide theoretical analyses, in a cue of marginalization approximation from in- to cross-domain scenarios, to uncover that the vanilla HLR suffers from an underlying mismatch between refined Gaussian conditional and hyperprior distributions, leading to deteriorated joint probability approximation of marginal distribution with increased rate consumption. To remedy this issue, we introduce a simple Bayesian approximation-endowed \textit{distribution regularization} to encourage learning a better joint probability approximation in a plug-and-play manner. Extensive experiments on six in- and cross-domain datasets demonstrate that our proposed method not only improves the R-D performance compared with other latent refinement counterparts, but also can be flexibly integrated into existing TTA-IC methods with incremental benefits.
Osteoarthritis is a degenerative condition affecting bones and cartilage, often leading to osteophyte formation, bone density loss, and joint space narrowing. Treatment options to restore normal joint function vary depending on the severity of the condition. This work introduces an innovative deep-learning framework processing shoulder CT scans. It features the semantic segmentation of the proximal humerus and scapula, the 3D reconstruction of bone surfaces, the identification of the glenohumeral (GH) joint region, and the staging of three common osteoarthritic-related pathologies: osteophyte formation (OS), GH space reduction (JS), and humeroscapular alignment (HSA). The pipeline comprises two cascaded CNN architectures: 3D CEL-UNet for segmentation and 3D Arthro-Net for threefold classification. A retrospective dataset of 571 CT scans featuring patients with various degrees of GH osteoarthritic-related pathologies was used to train, validate, and test the pipeline. Root mean squared error and Hausdorff distance median values for 3D reconstruction were 0.22mm and 1.48mm for the humerus and 0.24mm and 1.48mm for the scapula, outperforming state-of-the-art architectures and making it potentially suitable for a PSI-based shoulder arthroplasty preoperative plan context. The classification accuracy for OS, JS, and HSA consistently reached around 90% across all three categories. The computational time for the inference pipeline was less than 15s, showcasing the framework's efficiency and compatibility with orthopedic radiology practice. The outcomes represent a promising advancement toward the medical translation of artificial intelligence tools. This progress aims to streamline the preoperative planning pipeline delivering high-quality bone surfaces and supporting surgeons in selecting the most suitable surgical approach according to the unique patient joint conditions.
We propose a novel deep learning based method to design a coded waveform for integrated sensing and communication (ISAC) system based on orthogonal frequency-division multiplexing (OFDM). Our ultimate goal is to design a coded waveform, which is capable of providing satisfactory sensing performance of the target while maintaining high communication quality measured in terms of the bit error rate (BER). The proposed LISAC provides an improved waveform design with the assistance of deep neural networks for the encoding and decoding of the information bits. In particular, the transmitter, parameterized by a recurrent neural network (RNN), encodes the input bit sequence into the transmitted waveform for both sensing and communications. The receiver employs a RNN-based decoder to decode the information bits while the transmitter senses the target via maximum likelihood detection. We optimize the system considering both the communication and sensing performance. Simulation results show that the proposed LISAC waveform achieves a better trade-off curve compared to existing alternatives.
Multivariate Time series forecasting is crucial in domains such as transportation, meteorology, and finance, especially for predicting extreme weather events. State-of-the-art methods predominantly rely on Transformer architectures, which utilize attention mechanisms to capture temporal dependencies. However, these methods are hindered by quadratic time complexity, limiting the model's scalability with respect to input sequence length. This significantly restricts their practicality in the real world. Mamba, based on state space models (SSM), provides a solution with linear time complexity, increasing the potential for efficient forecasting of sequential data. In this study, we propose UmambaTSF, a novel long-term time series forecasting framework that integrates multi-scale feature extraction capabilities of U-shaped encoder-decoder multilayer perceptrons (MLP) with Mamba's long sequence representation. To improve performance and efficiency, the Mamba blocks introduced in the framework adopt a refined residual structure and adaptable design, enabling the capture of unique temporal signals and flexible channel processing. In the experiments, UmambaTSF achieves state-of-the-art performance and excellent generality on widely used benchmark datasets while maintaining linear time complexity and low memory consumption.
A proper scene representation is central to the pursuit of spatial intelligence where agents can robustly reconstruct and efficiently understand 3D scenes. A scene representation is either metric, such as landmark maps in 3D reconstruction, 3D bounding boxes in object detection, or voxel grids in occupancy prediction, or topological, such as pose graphs with loop closures in SLAM or visibility graphs in SfM. In this work, we propose to build Multiview Scene Graphs (MSG) from unposed images, representing a scene topologically with interconnected place and object nodes. The task of building MSG is challenging for existing representation learning methods since it needs to jointly address both visual place recognition, object detection, and object association from images with limited fields of view and potentially large viewpoint changes. To evaluate any method tackling this task, we developed an MSG dataset and annotation based on a public 3D dataset. We also propose an evaluation metric based on the intersection-over-union score of MSG edges. Moreover, we develop a novel baseline method built on mainstream pretrained vision models, combining visual place recognition and object association into one Transformer decoder architecture. Experiments demonstrate that our method has superior performance compared to existing relevant baselines.
Estimating depth from a single 2D image is a challenging task due to the lack of stereo or multi-view data, which are typically required for depth perception. In state-of-the-art architectures, the main challenge is to efficiently capture complex objects and fine-grained details, which are often difficult to predict. This paper introduces a novel deep learning-based approach using an enhanced encoder-decoder architecture, where the Inception-ResNet-v2 model serves as the encoder. This is the first instance of utilizing Inception-ResNet-v2 as an encoder for monocular depth estimation, demonstrating improved performance over previous models. It incorporates multi-scale feature extraction to enhance depth prediction accuracy across various object sizes and distances. We propose a composite loss function comprising depth loss, gradient edge loss, and Structural Similarity Index Measure (SSIM) loss, with fine-tuned weights to optimize the weighted sum, ensuring a balance across different aspects of depth estimation. Experimental results on the KITTI dataset show that our model achieves a significantly faster inference time of 0.019 seconds, outperforming vision transformers in efficiency while maintaining good accuracy. On the NYU Depth V2 dataset, the model establishes state-of-the-art performance, with an Absolute Relative Error (ARE) of 0.064, a Root Mean Square Error (RMSE) of 0.228, and an accuracy of 89.3% for $\delta$ < 1.25. These metrics demonstrate that our model can accurately and efficiently predict depth even in challenging scenarios, providing a practical solution for real-time applications.
Remote sensing image change detection (RSCD) is crucial for monitoring dynamic surface changes, with applications ranging from environmental monitoring to disaster assessment. While traditional CNN-based methods have improved detection accuracy, they often suffer from high computational complexity and large parameter counts, limiting their use in resource-constrained environments. To address these challenges, we propose a Lightweight remote sensing Change Detection Network (LCD-Net in short) that reduces model size and computational cost while maintaining high detection performance. LCD-Net employs MobileNetV2 as the encoder to efficiently extract features from bitemporal images. A Temporal Interaction and Fusion Module (TIF) enhances the interaction between bitemporal features, improving temporal context awareness. Additionally, the Feature Fusion Module (FFM) aggregates multiscale features to better capture subtle changes while suppressing background noise. The Gated Mechanism Module (GMM) in the decoder further enhances feature learning by dynamically adjusting channel weights, emphasizing key change regions. Experiments on LEVIR-CD+, SYSU, and S2Looking datasets show that LCD-Net achieves competitive performance with just 2.56M parameters and 4.45G FLOPs, making it well-suited for real-time applications in resource-limited settings. The code is available at https://github.com/WenyuLiu6/LCD-Net.
Open-vocabulary segmentation (OVS) has gained attention for its ability to recognize a broader range of classes. However, OVS models show significant performance drops when applied to unseen domains beyond the previous training dataset. Fine-tuning these models on new datasets can improve performance, but often leads to the catastrophic forgetting of previously learned knowledge. To address this issue, we propose a method that allows OVS models to learn information from new domains while preserving prior knowledge. Our approach begins by evaluating the input sample's proximity to multiple domains, using precomputed multivariate normal distributions for each domain. Based on this prediction, we dynamically interpolate between the weights of the pre-trained decoder and the fine-tuned decoders. Extensive experiments demonstrate that this approach allows OVS models to adapt to new domains while maintaining performance on the previous training dataset. The source code is available at https://github.com/dongjunhwang/dwi.
This paper presents CleanUMamba, a time-domain neural network architecture designed for real-time causal audio denoising directly applied to raw waveforms. CleanUMamba leverages a U-Net encoder-decoder structure, incorporating the Mamba state-space model in the bottleneck layer. By replacing conventional self-attention and LSTM mechanisms with Mamba, our architecture offers superior denoising performance while maintaining a constant memory footprint, enabling streaming operation. To enhance efficiency, we applied structured channel pruning, achieving an 8X reduction in model size without compromising audio quality. Our model demonstrates strong results in the Interspeech 2020 Deep Noise Suppression challenge. Specifically, CleanUMamba achieves a PESQ score of 2.42 and STOI of 95.1% with only 442K parameters and 468M MACs, matching or outperforming larger models in real-time performance. Code will be available at: https://github.com/lab-emi/CleanUMamba
Although existing variational graph autoencoders (VGAEs) have been widely used for modeling and generating graph-structured data, most of them are still not flexible enough to approximate the sparse and skewed latent node representations, especially those of document relational networks (DRNs) with discrete observations. To analyze a collection of interconnected documents, a typical branch of Bayesian models, specifically relational topic models (RTMs), has proven their efficacy in describing both link structures and document contents of DRNs, which motives us to incorporate RTMs with existing VGAEs to alleviate their potential issues when modeling the generation of DRNs. In this paper, moving beyond the sophisticated approximate assumptions of traditional RTMs, we develop a graph Poisson factor analysis (GPFA), which provides analytic conditional posteriors to improve the inference accuracy, and extend GPFA to a multi-stochastic-layer version named graph Poisson gamma belief network (GPGBN) to capture the hierarchical document relationships at multiple semantic levels. Then, taking GPGBN as the decoder, we combine it with various Weibull-based graph inference networks, resulting in two variants of Weibull graph auto-encoder (WGAE), equipped with model inference algorithms. Experimental results demonstrate that our models can extract high-quality hierarchical latent document representations and achieve promising performance on various graph analytic tasks.
Video Paragraph Captioning (VPC) aims to generate paragraph captions that summarises key events within a video. Despite recent advancements, challenges persist, notably in effectively utilising multimodal signals inherent in videos and addressing the long-tail distribution of words. The paper introduces a novel multimodal integrated caption generation framework for VPC that leverages information from various modalities and external knowledge bases. Our framework constructs two graphs: a 'video-specific' temporal graph capturing major events and interactions between multimodal information and commonsense knowledge, and a 'theme graph' representing correlations between words of a specific theme. These graphs serve as input for a transformer network with a shared encoder-decoder architecture. We also introduce a node selection module to enhance decoding efficiency by selecting the most relevant nodes from the graphs. Our results demonstrate superior performance across benchmark datasets.
Spatial awareness is key to enable embodied multimodal AI systems. Yet, without vast amounts of spatial supervision, current Visual Language Models (VLMs) struggle at this task. In this paper, we introduce LynX, a framework that equips pretrained VLMs with visual grounding ability without forgetting their existing image and language understanding skills. To this end, we propose a Dual Mixture of Experts module that modifies only the decoder layer of the language model, using one frozen Mixture of Experts (MoE) pre-trained on image and language understanding and another learnable MoE for new grounding capabilities. This allows the VLM to retain previously learned knowledge and skills, while acquiring what is missing. To train the model effectively, we generate a high-quality synthetic dataset we call SCouT, which mimics human reasoning in visual grounding. This dataset provides rich supervision signals, describing a step-by-step multimodal reasoning process, thereby simplifying the task of visual grounding. We evaluate LynX on several object detection and visual grounding datasets, demonstrating strong performance in object detection, zero-shot localization and grounded reasoning while maintaining its original image and language understanding capabilities on seven standard benchmark datasets.
Pubic symphysis-fetal head segmentation in transperineal ultrasound images plays a critical role for the assessment of fetal head descent and progression. Existing transformer segmentation methods based on sparse attention mechanism use handcrafted static patterns, which leads to great differences in terms of segmentation performance on specific datasets. To address this issue, we introduce a dynamic, query-aware sparse attention mechanism for ultrasound image segmentation. Specifically, we propose a novel method, named BRAU-Net to solve the pubic symphysis-fetal head segmentation task in this paper. The method adopts a U-Net-like encoder-decoder architecture with bi-level routing attention and skip connections, which effectively learns local-global semantic information. In addition, we propose an inverted bottleneck patch expanding (IBPE) module to reduce information loss while performing up-sampling operations. The proposed BRAU-Net is evaluated on FH-PS-AoP and HC18 datasets. The results demonstrate that our method could achieve excellent segmentation results. The code is available on GitHub.
This paper proposes to learn generative priors from the motion patterns instead of video contents for generative video compression. The priors are derived from small motion dynamics in common scenes such as swinging trees in the wind and floating boat on the sea. Utilizing such compact motion priors, a novel generative scene dynamics compression framework is built to realize ultra-low bit-rate communication and high-quality reconstruction for diverse scene contents. At the encoder side, motion priors are characterized into compact representations in a dense-to-sparse manner. At the decoder side, the decoded motion priors serve as the trajectory hints for scene dynamics reconstruction via a diffusion-based flow-driven generator. The experimental results illustrate that the proposed method can achieve superior rate-distortion performance and outperform the state-of-the-art conventional video codec Versatile Video Coding (VVC) on scene dynamics sequences. The project page can be found at https://github.com/xyzysz/GNVDC.
Segment Anything Model (SAM) has gained significant recognition in the field of semantic segmentation due to its versatile capabilities and impressive performance. Despite its success, SAM faces two primary limitations: (1) it relies heavily on meticulous human-provided prompts like key points, bounding boxes or text messages, which is labor-intensive; (2) the mask decoder's feature representation is sometimes inaccurate, as it solely employs dot product operations at the end of mask decoder, which inadequately captures the necessary correlations for precise segmentation. Current solutions to these problems such as fine-tuning SAM often require retraining a large number of parameters, which needs huge amount of time and computing resources. To address these limitations, we propose an automated prompting and mask calibration method called AM-SAM based on a bi-level optimization framework. Our approach automatically generates prompts for an input image, eliminating the need for human involvement with a good performance in early training epochs, achieving faster convergence. Additionally, we freeze the main part of SAM, and modify the mask decoder with Low-Rank Adaptation (LoRA), enhancing the mask decoder's feature representation by incorporating advanced techniques that go beyond simple dot product operations to more accurately capture and utilize feature correlations. Our experimental results demonstrate that AM-SAM achieves significantly accurate segmentation, matching or exceeding the effectiveness of human-generated and default prompts. Notably, on the body segmentation dataset, our method yields a 5% higher dice score with a 4-example few-shot training set compared to the SOTA method, underscoring its superiority in semantic segmentation tasks.