Medical image segmentation is increasingly reliant on deep learning techniques, yet the promising performance often come with high annotation costs. This paper introduces Weak-Mamba-UNet, an innovative weakly-supervised learning (WSL) framework that leverages the capabilities of Convolutional Neural Network (CNN), Vision Transformer (ViT), and the cutting-edge Visual Mamba (VMamba) architecture for medical image segmentation, especially when dealing with scribble-based annotations. The proposed WSL strategy incorporates three distinct architecture but same symmetrical encoder-decoder networks: a CNN-based UNet for detailed local feature extraction, a Swin Transformer-based SwinUNet for comprehensive global context understanding, and a VMamba-based Mamba-UNet for efficient long-range dependency modeling. The key concept of this framework is a collaborative and cross-supervisory mechanism that employs pseudo labels to facilitate iterative learning and refinement across the networks. The effectiveness of Weak-Mamba-UNet is validated on a publicly available MRI cardiac segmentation dataset with processed scribble annotations, where it surpasses the performance of a similar WSL framework utilizing only UNet or SwinUNet. This highlights its potential in scenarios with sparse or imprecise annotations. The source code is made publicly accessible.
Cross-technology communication (CTC) is a promising technique that enables direct communications among incompatible wireless technologies without needing hardware modification. However, it has not been widely adopted in real-world applications due to its inefficiency and unreliability. To address this issue, this paper proposes a deep joint source-channel coding (DJSCC) scheme to enable efficient and reliable CTC. The proposed scheme builds a neural-network-based encoder and decoder at the sender side and the receiver side, respectively, to achieve two critical tasks simultaneously: 1) compressing the messages to the point where only their essential semantic meanings are preserved; 2) ensuring the robustness of the semantic meanings when they are transmitted across incompatible technologies. The scheme incorporates existing CTC coding algorithms as domain knowledge to guide the encoder-decoder pair to learn the characteristics of CTC links better. Moreover, the scheme constructs shared semantic knowledge for the encoder and decoder, allowing semantic meanings to be converted into very few bits for cross-technology transmissions, thus further improving the efficiency of CTC. Extensive simulations verify that the proposed scheme can reduce the transmission overhead by up to 97.63\% and increase the structural similarity index measure by up to 734.78%, compared with the state-of-the-art CTC scheme.
Energy efficiency for video communications is essential for mobile devices with a limited battery capacity. Therefore, hardware decoder implementations are commonly used to significantly reduce the energetic load of video playback. The energy consumption of such a hardware implementation largely depends on a previously published specification of a video coding standard that defines which coding tools and methods are included. However, during the standardization of a video coding standard, the energy demand of a hardware implementation is unknown. Hence, the hardware complexity of coding tools is judged subjectively by experts from the field of hardware programming without using standardized assessment procedures. To solve this problem, we propose a method that accurately models the energy demand of existing hardware decoders with an average error of 1.79% by exploiting information from software decoder profiling. Motivated by the low estimation error, we propose a hardware decoding energy metric that can predict and estimate the energy demand of an unknown hardware implementation using information from existing hardware decoder implementations and available software implementations of the future video decoder. By using multiple video coding standards for model training, we can predict the relative energy demand of an unknown hardware decoder with a minimum error of 4.54% without using the corresponding hardware decoder for training.
Action quality assessment (AQA) is to assess how well an action is performed. Previous works perform modelling by only the use of visual information, ignoring audio information. We argue that although AQA is highly dependent on visual information, the audio is useful complementary information for improving the score regression accuracy, especially for sports with background music, such as figure skating and rhythmic gymnastics. To leverage multimodal information for AQA, i.e., RGB, optical flow and audio information, we propose a Progressive Adaptive Multimodal Fusion Network (PAMFN) that separately models modality-specific information and mixed-modality information. Our model consists of with three modality-specific branches that independently explore modality-specific information and a mixed-modality branch that progressively aggregates the modality-specific information from the modality-specific branches. To build the bridge between modality-specific branches and the mixed-modality branch, three novel modules are proposed. First, a Modality-specific Feature Decoder module is designed to selectively transfer modality-specific information to the mixed-modality branch. Second, when exploring the interaction between modality-specific information, we argue that using an invariant multimodal fusion policy may lead to suboptimal results, so as to take the potential diversity in different parts of an action into consideration. Therefore, an Adaptive Fusion Module is proposed to learn adaptive multimodal fusion policies in different parts of an action. This module consists of several FusionNets for exploring different multimodal fusion strategies and a PolicyNet for deciding which FusionNets are enabled. Third, a module called Cross-modal Feature Decoder is designed to transfer cross-modal features generated by Adaptive Fusion Module to the mixed-modality branch.
Adversarial attacks represent a substantial challenge in Natural Language Processing (NLP). This study undertakes a systematic exploration of this challenge in two distinct phases: vulnerability evaluation and resilience enhancement of Transformer-based models under adversarial attacks. In the evaluation phase, we assess the susceptibility of three Transformer configurations, encoder-decoder, encoder-only, and decoder-only setups, to adversarial attacks of escalating complexity across datasets containing offensive language and misinformation. Encoder-only models manifest a 14% and 21% performance drop in offensive language detection and misinformation detection tasks, respectively. Decoder-only models register a 16% decrease in both tasks, while encoder-decoder models exhibit a maximum performance drop of 14% and 26% in the respective tasks. The resilience-enhancement phase employs adversarial training, integrating pre-camouflaged and dynamically altered data. This approach effectively reduces the performance drop in encoder-only models to an average of 5% in offensive language detection and 2% in misinformation detection tasks. Decoder-only models, occasionally exceeding original performance, limit the performance drop to 7% and 2% in the respective tasks. Although not surpassing the original performance, Encoder-decoder models can reduce the drop to an average of 6% and 2% respectively. Results suggest a trade-off between performance and robustness, with some models maintaining similar performance while gaining robustness. Our study and adversarial training techniques have been incorporated into an open-source tool for generating camouflaged datasets. However, methodology effectiveness depends on the specific camouflage technique and data encountered, emphasizing the need for continued exploration.
Do pretrained language models have knowledge regarding the surface information of tokens? We examined the surface information stored in word or subword embeddings acquired by pretrained language models from the perspectives of token length, substrings, and token constitution. Additionally, we evaluated the ability of models to generate knowledge regarding token surfaces. We focused on 12 pretrained language models that were mainly trained on English and Japanese corpora. Experimental results demonstrate that pretrained language models have knowledge regarding token length and substrings but not token constitution. Additionally, the results imply that there is a bottleneck on the decoder side in terms of effectively utilizing acquired knowledge.
This study explores the potential of fine-tuning foundational English Large Language Models (LLMs) for generating Polish text. The first step involves Language Adaptive Pre-training (LAPT) on a high-quality dataset of 3.11 GB, consisting of 276 million Polish tokens. The LAPT is followed by additional fine-tuning aimed at solving nine KLEJ challenges. Our trained model Curie-7B-v1 not only generates Polish text with the lowest perplexity of 3.02 among decoder-based Polish models but also closely rivals the performance of the best Polish encoder-decoder models with a less than 2% gap on 8 out of 9 tasks. Curie-7B-v1 used approximately 2-3% of a typical dataset size to learn Polish. The LAPT was completed in less than five days using a consumer GPU, highlighting the method's efficiency. The proficiency of the model in Polish was significantly enhanced, demonstrating the viability of this approach for adding new languages to existing LLMs by training just 1.2% of its parameters. To contribute to the community's collaborative progress, the model has been released as open-source.
Zero-shot text-to-speech (TTS) has gained significant attention due to its powerful voice cloning capabilities, requiring only a few seconds of unseen speaker voice prompts. However, all previous work has been developed for cloud-based systems. Taking autoregressive models as an example, although these approaches achieve high-fidelity voice cloning, they fall short in terms of inference speed, model size, and robustness. Therefore, we propose MobileSpeech, which is a fast, lightweight, and robust zero-shot text-to-speech system based on mobile devices for the first time. Specifically: 1) leveraging discrete codec, we design a parallel speech mask decoder module called SMD, which incorporates hierarchical information from the speech codec and weight mechanisms across different codec layers during the generation process. Moreover, to bridge the gap between text and speech, we introduce a high-level probabilistic mask that simulates the progression of information flow from less to more during speech generation. 2) For speaker prompts, we extract fine-grained prompt duration from the prompt speech and incorporate text, prompt speech by cross attention in SMD. We demonstrate the effectiveness of MobileSpeech on multilingual datasets at different levels, achieving state-of-the-art results in terms of generating speed and speech quality. MobileSpeech achieves RTF of 0.09 on a single A100 GPU and we have successfully deployed MobileSpeech on mobile devices. Audio samples are available at \url{https://mobilespeech.github.io/} .
Energy and compression efficiency are two essential parts of modern video decoder implementations that have to be considered. This work comprehensively studies the following six video coding formats regarding compression and decoding energy efficiency: AVC, VP9, HEVC, AV1, VVC, and AVM. We first evaluate the energy demand of reference and optimized software decoder implementations. Furthermore, we consider the influence of the usage of SIMD instructions on those decoder implementations. We find that AV1 is a sweet spot for optimized software decoder implementations with an additional energy demand of 16.55% and bitrate savings of -43.95% compared to VP9. We furthermore evaluate the hardware decoding energy demand of four video coding formats. Thereby, we show that AV1 has energy demand increases by 117.50% compared to VP9. For HEVC, we found a sweet spot in terms of energy demand with an increase of 6.06% with respect to VP9. Relative to their optimized software counterparts, hardware video decoders reduce the energy consumption to less than 9% compared to software decoders.
Non-autoregressive automatic speech recognition (NASR) models have gained attention due to their parallelism and fast inference. The encoder-based NASR, e.g. connectionist temporal classification (CTC), can be initialized from the speech foundation models (SFM) but does not account for any dependencies among intermediate tokens. The encoder-decoder-based NASR, like CTC alignment-based single-step non-autoregressive transformer (CASS-NAT), can mitigate the dependency problem but is not able to efficiently integrate SFM. Inspired by the success of recent work of speech-text joint pre-training with a shared transformer encoder, we propose a new encoder-based NASR, UniEnc-CASSNAT, to combine the advantages of CTC and CASS-NAT. UniEnc-CASSNAT consists of only an encoder as the major module, which can be the SFM. The encoder plays the role of both the CASS-NAT encoder and decoder by two forward passes. The first pass of the encoder accepts the speech signal as input, while the concatenation of the speech signal and the token-level acoustic embedding is used as the input for the second pass. Examined on the Librispeech 100h, MyST, and Aishell1 datasets, the proposed UniEnc-CASSNAT achieves state-of-the-art NASR results and is better or comparable to CASS-NAT with only an encoder and hence, fewer model parameters. Our codes are publicly available.
The advent of deep-learning-based registration networks has addressed the time-consuming challenge in traditional iterative methods.However, the potential of current registration networks for comprehensively capturing spatial relationships has not been fully explored, leading to inadequate performance in large-deformation image registration.The pure convolutional neural networks (CNNs) neglect feature enhancement, while current Transformer-based networks are susceptible to information redundancy.To alleviate these issues, we propose a pyramid attention network (PAN) for deformable medical image registration.Specifically, the proposed PAN incorporates a dual-stream pyramid encoder with channel-wise attention to boost the feature representation.Moreover, a multi-head local attention Transformer is introduced as decoder to analyze motion patterns and generate deformation fields.Extensive experiments on two public brain magnetic resonance imaging (MRI) datasets and one abdominal MRI dataset demonstrate that our method achieves favorable registration performance, while outperforming several CNN-based and Transformer-based registration networks.Our code is publicly available at https://github.com/JuliusWang-7/PAN.
Diffusion models have achieved remarkable success in generating high quality image and video data. More recently, they have also been used for image compression with high perceptual quality. In this paper, we present a novel approach to extreme video compression leveraging the predictive power of diffusion-based generative models at the decoder. The conditional diffusion model takes several neural compressed frames and generates subsequent frames. When the reconstruction quality drops below the desired level, new frames are encoded to restart prediction. The entire video is sequentially encoded to achieve a visually pleasing reconstruction, considering perceptual quality metrics such as the learned perceptual image patch similarity (LPIPS) and the Frechet video distance (FVD), at bit rates as low as 0.02 bits per pixel (bpp). Experimental results demonstrate the effectiveness of the proposed scheme compared to standard codecs such as H.264 and H.265 in the low bpp regime. The results showcase the potential of exploiting the temporal relations in video data using generative models. Code is available at: https://github.com/ElesionKyrie/Extreme-Video-Compression-With-Prediction-Using-Pre-trainded-Diffusion-Models-
Dynamic scene understanding is one of the most conspicuous field of interest among computer vision community. In order to enhance dynamic scene understanding, pixel-wise segmentation with neural networks is widely accepted. The latest researches on pixel-wise segmentation combined semantic and motion information and produced good performance. In this work, we propose a state of art architecture of neural networks to accurately and efficiently get the moving object proposals (MOP). We first train an unsupervised convolutional neural network (UnFlow) to generate optical flow estimation. Then we render the output of optical flow net to a fully convolutional SegNet model. The main contribution of our work is (1) Fine-tuning the pretrained optical flow model on the brand new DAVIS Dataset; (2) Leveraging fully convolutional neural networks with Encoder-Decoder architecture to segment objects. We developed the codes with TensorFlow, and executed the training and evaluation processes on an AWS EC2 instance.
We present PECMAE, an interpretable model for music audio classification based on prototype learning. Our model is based on a previous method, APNet, which jointly learns an autoencoder and a prototypical network. Instead, we propose to decouple both training processes. This enables us to leverage existing self-supervised autoencoders pre-trained on much larger data (EnCodecMAE), providing representations with better generalization. APNet allows prototypes' reconstruction to waveforms for interpretability relying on the nearest training data samples. In contrast, we explore using a diffusion decoder that allows reconstruction without such dependency. We evaluate our method on datasets for music instrument classification (Medley-Solos-DB) and genre recognition (GTZAN and a larger in-house dataset), the latter being a more challenging task not addressed with prototypical networks before. We find that the prototype-based models preserve most of the performance achieved with the autoencoder embeddings, while the sonification of prototypes benefits understanding the behavior of the classifier.
We introduce a text-to-speech (TTS) model called BASE TTS, which stands for $\textbf{B}$ig $\textbf{A}$daptive $\textbf{S}$treamable TTS with $\textbf{E}$mergent abilities. BASE TTS is the largest TTS model to-date, trained on 100K hours of public domain speech data, achieving a new state-of-the-art in speech naturalness. It deploys a 1-billion-parameter autoregressive Transformer that converts raw texts into discrete codes ("speechcodes") followed by a convolution-based decoder which converts these speechcodes into waveforms in an incremental, streamable manner. Further, our speechcodes are built using a novel speech tokenization technique that features speaker ID disentanglement and compression with byte-pair encoding. Echoing the widely-reported "emergent abilities" of large language models when trained on increasing volume of data, we show that BASE TTS variants built with 10K+ hours and 500M+ parameters begin to demonstrate natural prosody on textually complex sentences. We design and share a specialized dataset to measure these emergent abilities for text-to-speech. We showcase state-of-the-art naturalness of BASE TTS by evaluating against baselines that include publicly available large-scale text-to-speech systems: YourTTS, Bark and TortoiseTTS. Audio samples generated by the model can be heard at https://amazon-ltts-paper.com/.
Sign language to text is a crucial technology that can break down communication barriers for individuals with hearing difficulties. We replicate and try to improve on a recently published study. We evaluate models using BLEU and rBLEU metrics to ensure translation quality. During our ablation study, we found that the model's performance is significantly influenced by optimizers, activation functions, and label smoothing. Further research aims to refine visual feature capturing, enhance decoder utilization, and integrate pre-trained decoders for better translation outcomes. Our source code is available to facilitate replication of our results and encourage future research.
The ability to generate sentiment-controlled feedback in response to multimodal inputs comprising text and images addresses a critical gap in human-computer interaction. This capability allows systems to provide empathetic, accurate, and engaging responses, with useful applications in education, healthcare, marketing, and customer service. To this end, we have constructed a large-scale Controllable Multimodal Feedback Synthesis (CMFeed) dataset and propose a controllable feedback synthesis system. The system features an encoder, decoder, and controllability block for textual and visual inputs. It extracts features using a transformer and Faster R-CNN networks, combining them to generate feedback. The CMFeed dataset includes images, texts, reactions to the posts, human comments with relevance scores, and reactions to these comments. These reactions train the model to produce feedback with specified sentiments, achieving a sentiment classification accuracy of 77.23\%, which is 18.82\% higher than the accuracy without controllability. The system also incorporates a similarity module for assessing feedback relevance through rank-based metrics and an interpretability technique to analyze the contributions of textual and visual features during feedback generation. Access to the CMFeed dataset and the system's code is available at https://github.com/MIntelligence-Group/CMFeed.
Large language models (LLMs) predominantly employ decoder-only transformer architectures, necessitating the retention of keys/values information for historical tokens to provide contextual information and avoid redundant computation. However, the substantial size and parameter volume of these LLMs require massive GPU memory. This memory demand increases with the length of the input text, leading to an urgent need for more efficient methods of information storage and processing. This study introduces Anchor-based LLMs (AnLLMs), which utilize an innovative anchor-based self-attention network (AnSAN) and also an anchor-based inference strategy. This approach enables LLMs to compress sequence information into an anchor token, reducing the keys/values cache and enhancing inference efficiency. Experiments on question-answering benchmarks reveal that AnLLMs maintain similar accuracy levels while achieving up to 99% keys/values cache reduction and up to 3.5 times faster inference. Despite a minor compromise in accuracy, the substantial enhancements of AnLLMs employing the AnSAN technique in resource utilization and computational efficiency underscore their potential for practical LLM applications.
In recent years, neural networks have been used to solve phase retrieval problems in imaging with superior accuracy and speed than traditional techniques, especially in the presence of noise. However, in the context of interferometric imaging, phase noise has been largely unaddressed by existing neural network architectures. Such noise arises naturally in an interferometer due to mechanical instabilities or atmospheric turbulence, limiting measurement acquisition times and posing a challenge in scenarios with limited light intensity, such as remote sensing. Here, we introduce a 3D-2D Phase Retrieval U-Net (PRUNe) that takes noisy and randomly phase-shifted interferograms as inputs, and outputs a single 2D phase image. A 3D downsampling convolutional encoder captures correlations within and between frames to produce a 2D latent space, which is upsampled by a 2D decoder into a phase image. We test our model against a state-of-the-art singular value decomposition algorithm and find PRUNe reconstructions consistently show more accurate and smooth reconstructions, with a x2.5 - 4 lower mean squared error at multiple signal-to-noise ratios for interferograms with low (< 1 photon/pixel) and high (~100 photons/pixel) signal intensity. Our model presents a faster and more accurate approach to perform phase retrieval in extremely low light intensity interferometry in presence of phase noise, and will find application in other multi-frame noisy imaging techniques.
Ring artifacts in X-ray micro-CT images are one of the primary causes of concern in their accurate visual interpretation and quantitative analysis. The geometry of X-ray micro-CT scanners is similar to the medical CT machines, except the sample is rotated with a stationary source and detector. The ring artifacts are caused by a defect or non-linear responses in detector pixels during the MicroCT data acquisition. Artifacts in MicroCT images can often be so severe that the images are no longer useful for further analysis. Therefore, it is essential to comprehend the causes of artifacts and potential solutions to maximize image quality. This article presents a convolution neural network (CNN)-based Deep Learning (DL) model inspired by UNet with a series of encoder and decoder units with skip connections for removal of ring artifacts. The proposed architecture has been evaluated using the Structural Similarity Index Measure (SSIM) and Mean Squared Error (MSE). Additionally, the results are compared with conventional filter-based non-ML techniques and are found to be better than the latter.