Data-oriented Dynamic Fine-tuning Parameter Selection Strategy for FISH Mask based Efficient Fine-tuning

In view of the huge number of parameters of Large language models (LLMs) , tuning all parameters is very costly, and accordingly fine-tuning specific parameters is more sensible. Most of parameter efficient fine-tuning (PEFT) concentrate on parameter selection strategies, such as additive method, selective method and reparametrization-based method. However, there are few methods that consider the impact of data samples on parameter selecting, such as Fish Mask based method. Fish Mask randomly choose a part of data samples and treat them equally during parameter selection, which is unable to dynamically select optimal parameters for inconstant data distributions. In this work, we adopt a data-oriented perspective, then proposing an IRD ($\mathrm{\underline I}$terative sample-parameter $\mathrm{\underline R}$ange $\mathrm{\underline D}$ecreasing) algorithm to search the best setting of sample-parameter pair for FISH Mask. In each iteration, by searching the set of samples and parameters with larger Fish information, IRD can find better sample-parameter pair in most scale. We demonstrate the effectiveness and rationality of proposed strategy by conducting experiments on GLUE benchmark. Experimental results show our strategy optimizes the parameter selection and achieves preferable performance.

Rich Semantic Knowledge Enhanced Large Language Models for Few-shot Chinese Spell Checking

Chinese Spell Checking (CSC) is a widely used technology, which plays a vital role in speech to text (STT) and optical character recognition (OCR). Most of the existing CSC approaches relying on BERT architecture achieve excellent performance. However, limited by the scale of the foundation model, BERT-based method does not work well in few-shot scenarios, showing certain limitations in practical applications. In this paper, we explore using an in-context learning method named RS-LLM (Rich Semantic based LLMs) to introduce large language models (LLMs) as the foundation model. Besides, we study the impact of introducing various Chinese rich semantic information in our framework. We found that by introducing a small number of specific Chinese rich semantic structures, LLMs achieve better performance than the BERT-based model on few-shot CSC task. Furthermore, we conduct experiments on multiple datasets, and the experimental results verified the superiority of our proposed framework.

Joint Learning for Scattered Point Cloud Understanding with Hierarchical Self-Distillation

Numerous point-cloud understanding techniques focus on whole entities and have succeeded in obtaining satisfactory results and limited sparsity tolerance. However, these methods are generally sensitive to incomplete point clouds that are scanned with flaws or large gaps. To address this issue, in this paper, we propose an end-to-end architecture that compensates for and identifies partial point clouds on the fly. First, we propose a cascaded solution that integrates both the upstream and downstream networks simultaneously, allowing the task-oriented downstream to identify the points generated by the completion-oriented upstream. These two streams complement each other, resulting in improved performance for both completion and downstream-dependent tasks. Second, to explicitly understand the predicted points' pattern, we introduce hierarchical self-distillation (HSD), which can be applied to arbitrary hierarchy-based point cloud methods. HSD ensures that the deepest classifier with a larger perceptual field and longer code length provides additional regularization to intermediate ones rather than simply aggregating the multi-scale features, and therefore maximizing the mutual information between a teacher and students. We show the advantage of the self-distillation process in the hyperspaces based on the information bottleneck principle. On the classification task, our proposed method performs competitively on the synthetic dataset and achieves superior results on the challenging real-world benchmark when compared to the state-of-the-art models. Additional experiments also demonstrate the superior performance and generality of our framework on the part segmentation task.

Modality-Agnostic Learning for Medical Image Segmentation Using Multi-modality Self-distillation

Medical image segmentation of tumors and organs at risk is a time-consuming yet critical process in the clinic that utilizes multi-modality imaging (e.g, different acquisitions, data types, and sequences) to increase segmentation precision. In this paper, we propose a novel framework, Modality-Agnostic learning through Multi-modality Self-dist-illation (MAG-MS), to investigate the impact of input modalities on medical image segmentation. MAG-MS distills knowledge from the fusion of multiple modalities and applies it to enhance representation learning for individual modalities. Thus, it provides a versatile and efficient approach to handle limited modalities during testing. Our extensive experiments on benchmark datasets demonstrate the high efficiency of MAG-MS and its superior segmentation performance than current state-of-the-art methods. Furthermore, using MAG-MS, we provide valuable insight and guidance on selecting input modalities for medical image segmentation tasks.

Dual self-distillation of U-shaped networks for 3D medical image segmentation

U-shaped networks and its variants have demonstrated exceptional results for medical image segmentation. In this paper, we propose a novel dual self-distillation (DSD) framework for U-shaped networks for 3D medical image segmentation. DSD distills knowledge from the ground-truth segmentation labels to the decoder layers and also between the encoder and decoder layers of a single U-shaped network. DSD is a generalized training strategy that could be attached to the backbone architecture of any U-shaped network to further improve its segmentation performance. We attached DSD on two state-of-the-art U-shaped backbones, and extensive experiments on two public 3D medical image segmentation datasets (cardiac substructure and brain tumor) demonstrated significant improvement over those backbones. On average, after attaching DSD to the U-shaped backbones, we observed an improvement of 4.25% and 3.15% in Dice similarity score for cardiac substructure and brain tumor segmentation respectively.

Spatial-Temporal Synchronous Graph Transformer network (STSGT) for COVID-19 forecasting

COVID-19 has become a matter of serious concern over the last few years. It has adversely affected numerous people around the globe and has led to the loss of billions of dollars of business capital. In this paper, we propose a novel Spatial-Temporal Synchronous Graph Transformer network (STSGT) to capture the complex spatial and temporal dependency of the COVID-19 time series data and forecast the future status of an evolving pandemic. The layers of STSGT combine the graph convolution network (GCN) with the self-attention mechanism of transformers on a synchronous spatial-temporal graph to capture the dynamically changing pattern of the COVID time series. The spatial-temporal synchronous graph simultaneously captures the spatial and temporal dependencies between the vertices of the graph at a given and subsequent time-steps, which helps capture the heterogeneity in the time series and improve the forecasting accuracy. Our extensive experiments on two publicly available real-world COVID-19 time series datasets demonstrate that STSGT significantly outperforms state-of-the-art algorithms that were designed for spatial-temporal forecasting tasks. Specifically, on average over a 12-day horizon, we observe a potential improvement of 12.19% and 3.42% in Mean Absolute Error(MAE) over the next best algorithm while forecasting the daily infected and death cases respectively for the 50 states of US and Washington, D.C. Additionally, STSGT also outperformed others when forecasting the daily infected cases at the state level, e.g., for all the counties in the State of Michigan. The code and models are publicly available at https://github.com/soumbane/STSGT.

Automated Identification of Toxic Code Reviews: How Far Can We Go?

Toxic conversations during software development interactions may have serious repercussions on a Free and Open Source Software (FOSS) development project. For example, victims of toxic conversations may become afraid to express themselves, therefore get demotivated, and may eventually leave the project. Automated filtering of toxic conversations may help a FOSS community to maintain healthy interactions among its members. However, off-the-shelf toxicity detectors perform poorly on Software Engineering (SE) dataset, such as one curated from code review comments. To encounter this challenge, we present ToxiCR, a supervised learning-based toxicity identification tool for code review interactions. ToxiCR includes a choice to select one of the ten supervised learning algorithms, an option to select text vectorization techniques, five mandatory and three optional SE domain specific processing steps, and a large scale labeled dataset of 19,571 code review comments. With our rigorous evaluation of the models with various combinations of preprocessing steps and vectorization techniques, we have identified the best combination for our dataset that boosts 95.8% accuracy and 88.9% F1 score. ToxiCR significantly outperforms existing toxicity detectors on our dataset. We have released our dataset, pretrained models, evaluation results, and source code publicly available at: https://github.com/WSU-SEAL/ToxiCR.

Learning Pruned Structure and Weights Simultaneously from Scratch: an Attention based Approach

As a deep learning model typically contains millions of trainable weights, there has been a growing demand for a more efficient network structure with reduced storage space and improved run-time efficiency. Pruning is one of the most popular network compression techniques. In this paper, we propose a novel unstructured pruning pipeline, Attention-based Simultaneous sparse structure and Weight Learning (ASWL). Unlike traditional channel-wise or weight-wise attention mechanism, ASWL proposed an efficient algorithm to calculate the pruning ratio through layer-wise attention for each layer, and both weights for the dense network and the sparse network are tracked so that the pruned structure is simultaneously learned from randomly initialized weights. Our experiments on MNIST, Cifar10, and ImageNet show that ASWL achieves superior pruning results in terms of accuracy, pruning ratio and operating efficiency when compared with state-of-the-art network pruning methods.

"Zero Shot" Point Cloud Upsampling

Point cloud upsampling using deep learning has been paid various efforts in the past few years. Recent supervised deep learning methods are restricted to the size of training data and is limited in terms of covering all shapes of point clouds. Besides, the acquisition of such amount of data is unrealistic, and the network generally performs less powerful than expected on unseen records. In this paper, we present an unsupervised approach to upsample point clouds internally referred as "Zero Shot" Point Cloud Upsampling (ZSPU) at holistic level. Our approach is solely based on the internal information provided by a particular point cloud without patching in both self-training and testing phases. This single-stream design significantly reduces the training time of the upsampling task, by learning the relation between low-resolution (LR) point clouds and their high (original) resolution (HR) counterparts. This association will provide super-resolution (SR) outputs when original point clouds are loaded as input. We demonstrate competitive performance on benchmark point cloud datasets when compared to other upsampling methods. Furthermore, ZSPU achieves superior qualitative results on shapes with complex local details or high curvatures.

SA-GAN: Structure-Aware Generative Adversarial Network for Shape-Preserving Synthetic CT Generation

In medical image synthesis, model training could be challenging due to the inconsistencies between images of different modalities even with the same patient, typically caused by internal status/tissue changes as different modalities are usually obtained at a different time. This paper proposes a novel deep learning method, Structure-aware Generative Adversarial Network (SA-GAN), that preserves the shapes and locations of in-consistent structures when generating medical images. SA-GAN is employed to generate synthetic computed tomography (synCT) images from magnetic resonance imaging (MRI) with two parallel streams: the global stream translates the input from the MRI to the CT domain while the local stream automatically segments the inconsistent organs, maintains their locations and shapes in MRI, and translates the organ intensities to CT. Through extensive experiments on a pelvic dataset, we demonstrate that SA-GAN provides clinically acceptable accuracy on both synCTs and organ segmentation and supports MR-only treatment planning in disease sites with internal organ status changes.