Prediction Task Research Articles

Constant Companionship Without Disturbances: Enhancing Transparency to Improve Automated Tasks in Urban Rail Transit Driving

Enhancing transparency through interface design is an effective method for improving driving safety while reducing driver workloads, potentially fostering human–machine collaboration. However, to ensure system usability and safety, operator psychological factors and operational performance must be well balanced. This study investigates how the introduction of transparency design into urban rail transit driving tasks influences drivers’ situational awareness (SA), trust in automation (TiA), sense of agency (SoA), workload, operational performance, and visual behavior. Three transparency driver–machine interface (DMI) information conditions were evaluated: DMI1, which provided continuous feedback on vehicle operating status and actions; DMI1+2, which added inferential explanations; and DMI1+2+3, which further incorporated proactive predictions. Results from simulated driving experiments with 32 participants indicated that an appropriate level of transparency significantly enhanced TiA and SoA, thereby yielding the greatest acceptance. High transparency significantly aided in predictable takeover tasks but affected gains in TiA and SoA, increased workload, and disrupted perception-level SA. Compared with previous research findings, this study indicates the presence of a disparity in transparency needs for low-workload tasks. Therefore, caution should be exercised when introducing high-transparency designs in urban rail transit driving tasks. Nonetheless, an appropriate transparency interface design can enhance the driving experience.

EpiGePT: a pretrained transformer-based language model for context-specific human epigenomics

The inherent similarities between natural language and biological sequences have inspired the use of large language models in genomics, but current models struggle to incorporate chromatin interactions or predict in unseen cellular contexts. To address this, we propose EpiGePT, a transformer-based model designed for predicting context-specific human epigenomic signals. By incorporating transcription factor activities and 3D genome interactions, EpiGePT outperforms existing methods in epigenomic signal prediction tasks, especially in cell-type-specific long-range interaction predictions and genetic variant impacts, advancing our understanding of gene regulation. A free online prediction service is available at http://health.tsinghua.edu.cn/epigept.

Robust Automated Mouse Micro-CT Segmentation Using Swin UNEt TRansformers

Image-guided mouse irradiation is essential to understand interventions involving radiation prior to human studies. Our objective is to employ Swin UNEt TRansformers (Swin UNETR) to segment native micro-CT and contrast-enhanced micro-CT scans and benchmark the results against 3D no-new-Net (nnU-Net). Swin UNETR reformulates mouse organ segmentation as a sequence-to-sequence prediction task using a hierarchical Swin Transformer encoder to extract features at five resolution levels, and it connects to a Fully Convolutional Neural Network (FCNN)-based decoder via skip connections. The models were trained and evaluated on open datasets, with data separation based on individual mice. Further evaluation on an external mouse dataset acquired on a different micro-CT with lower kVp and higher imaging noise was also employed to assess model robustness and generalizability. The results indicate that Swin UNETR consistently outperforms nnU-Net and AIMOS in terms of the average dice similarity coefficient (DSC) and the Hausdorff distance (HD95p), except in two mice for intestine contouring. This superior performance is especially evident in the external dataset, confirming the model’s robustness to variations in imaging conditions, including noise and quality, and thereby positioning Swin UNETR as a highly generalizable and efficient tool for automated contouring in pre-clinical workflows.

Toward Closing the Loop in Image-to-Image Conversion in Radiotherapy: A Quality Control Tool to Predict Synthetic Computed Tomography Hounsfield Unit Accuracy

In recent years, synthetic Computed Tomography (CT) images generated from Magnetic Resonance (MR) or Cone Beam Computed Tomography (CBCT) acquisitions have been shown to be comparable to real CT images in terms of dose computation for radiotherapy simulation. However, until now, there has been no independent strategy to assess the quality of each synthetic image in the absence of ground truth. In this work, we propose a Deep Learning (DL)-based framework to predict the accuracy of synthetic CT in terms of Mean Absolute Error (MAE) without the need for a ground truth (GT). The proposed algorithm generates a volumetric map as an output, informing clinicians of the predicted MAE slice-by-slice. A cascading multi-model architecture was used to deal with the complexity of the MAE prediction task. The workflow was trained and tested on two cohorts of head and neck cancer patients with different imaging modalities: 27 MR scans and 33 CBCT. The algorithm evaluation revealed an accurate HU prediction (a median absolute prediction deviation equal to 4 HU for CBCT-based synthetic CTs and 6 HU for MR-based synthetic CTs), with discrepancies that do not affect the clinical decisions made on the basis of the proposed estimation. The workflow exhibited no systematic error in MAE prediction. This work represents a proof of concept about the feasibility of synthetic CT evaluation in daily clinical practice, and it paves the way for future patient-specific quality assessment strategies.

An Evaluation of Zero-Cost Proxies - from Neural Architecture Performance Prediction to Model Robustness

AbstractZero-cost proxies are nowadays frequently studied and used to search for neural architectures. They show an impressive ability to predict the performance of architectures by making use of their untrained weights. These techniques allow for immense search speed-ups. So far the joint search for well performing and robust architectures has received much less attention in the field of NAS. Therefore, the main focus of zero-cost proxies is the clean accuracy of architectures, whereas the model robustness should play an evenly important part. In this paper, we analyze the ability of common zero-cost proxies to serve as performance predictors for robustness in the popular NAS-Bench-201 search space. We are interested in the single prediction task for robustness and the joint multi-objective of clean and robust accuracy. We further analyze the feature importance of the proxies and show that predicting the robustness makes the prediction task from existing zero-cost proxies more challenging. As a result, the joint consideration of several proxies becomes necessary to predict a model’s robustness while the clean accuracy can be regressed from a single such feature. Our code is available at https://github.com/jovitalukasik/zcp_eval.

Classification and prediction of heart disease using Machine Learning models: A promising approach for medical diagnosis

Background: heart disease is one of the leading causes of death worldwide, claiming 17.9 million lives. They are a major public health problem that affects people regardless of age or gender. Objective: This work aims to classify and predict heart disease using Machine Learning (ML) models such as Random Forest (RF), K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Naive Bayes (NB), Decision Tree (DT) and Logistic Regression (LR). Methods: We worked with the Cleveland dataset from Kaggle, consisting of 303 patient records and 14 attributes. This research was conducted in different stages, including model understanding, dataset analysis and cleaning, ML model training, and model performance evaluation. Results: The results showed that the RF and KNN models achieved the highest levels of performance and accuracy with 88.52%, surpassing the other models such as SVM, NB, and LR which obtained 86.89% accuracy, and DT with 78.69%. Conclusions: In conclusion, the RF and KNN models stand out over the other models for this type of prediction task.

SeqDPI: A 1D-CNN approach for predicting binding affinity of kinase inhibitors.

Predicting drug target binding affinity has huge relevance in Modern drug discovery and drug repositioning processes which assist doctors to come up with new drugs or even use the existing drugs for new target proteins. In silico models, using advanced deep learning techniques could further assist these prediction tasks by providing most prominent drug target pairs. Considering these factors, a deep learning based algorithmic framework is developed in this study to support drug target interaction prediction. The proposed SeqDPI model extract the relevant drug and protein features from the one dimensional Sequential representation of the dataset considered using optimized CNN networks that deploy convolutions on varying length of amino acid subsequence's to capture hidden pattern, the convolved drug- protein features obtained are then used as an input to L2 penalized feed forward neural network which matches the local residue patterns in protein classes with molecular fingerprints of drugs to predict the binding strength for all drug target pairs. The proposed model reduces the convolution strain typically encountered in existing in silico models that utilize complex 3D structures of drug protein datasets. The result shows that the SeqDPI model achieves a mean square error MSE of (0.167) across cross validation folds, outperforming baseline models such as KronRLS (0.406), Simboost (0.226), and DeepPS (0.214). Additionally, SeqDPI attains a high CI score of 0.9114 on the benchmark KIBA dataset, demonstrating its statistical significance and computational efficiency compared to existing methods. This gives the relevance and effectiveness of SeqDPI model in accurately predicting binding affinities while working with simpler one-dimensional data, making it a robust and computationally cost-effective solution for drug-target interaction prediction.

MTAF–DTA: multi-type attention fusion network for drug–target affinity prediction

BackgroundThe development of drug–target binding affinity (DTA) prediction tasks significantly drives the drug discovery process forward. Leveraging the rapid advancement of artificial intelligence, DTA prediction tasks have undergone a transformative shift from wet lab experimentation to machine learning-based prediction. This transition enables a more expedient exploration of potential interactions between drugs and targets, leading to substantial savings in time and funding resources. However, existing methods still face several challenges, such as drug information loss, lack of calculation of the contribution of each modality, and lack of simulation regarding the drug–target binding mechanisms.ResultsWe propose MTAF–DTA, a method for drug–target binding affinity prediction to solve the above problems. The drug representation module extracts three modalities of features from drugs and uses an attention mechanism to update their respective contribution weights. Additionally, we design a Spiral-Attention Block (SAB) as drug–target feature fusion module based on multi-type attention mechanisms, facilitating a triple fusion process between them. The SAB, to some extent, simulates the interactions between drugs and targets, thereby enabling outstanding performance in the DTA task. Our regression task on the Davis and KIBA datasets demonstrates the predictive capability of MTAF–DTA, with CI and MSE metrics showing respective improvements of 1.1% and 9.2% over the state-of-the-art (SOTA) method in the novel target settings. Furthermore, downstream tasks further validate MTAF–DTA’s superiority in DTA prediction.ConclusionsExperimental results and case study demonstrate the superior performance of our approach in DTA prediction tasks, showing its potential in practical applications such as drug discovery and disease treatment.

PREDICTION BASED ON MACHINE LEARNING OF TOOTH SENSITIVITY FOR IN-OFFICE DENTAL BLEACHING

To develop a supervised machine learning model to predict the occurrence and intensity of tooth sensitivity (TS) in patients undergoing in-office dental bleaching testing various algorithm models. Retrospective data from 458 patients were analyzed, including variables such as the occurrence and intensity of TS, basal tooth color, bleaching material characteristics (concentration and pH), intervention details (number and duration of applications), and patient age. Classification and regression models were evaluated using 5-fold cross-validation and assessed based on various performance parameters. For the predictive classification task (occurrence of TS), the developed models achieved a maximum area under the receiver operating characteristic curve (AUC) of 0.76 [0.62-0.88] on the test data, with an F1-score of 0.80 [0.71-0.87]. In cross-validation, the highest AUC reached 0.86 [0.84-0.88], and the highest F1-score was 0.78 [0.75-0.83]. For predicting TS intensity, the regression models demonstrated a minimum mean absolute error (MAE) of 1.76 [1.45-2.06] and a root mean square error (RMSE) of 2.38 [2.06-2.69] on the test set. During cross-validation, the lowest MAE was 1.84 [1.67-2.03], with an RMSE of 2.39 [2.20-2.58]. The supervised machine learning model for estimating the occurrence and intensity of TS in patients undergoing in-office bleaching demonstrated good predictive power. The Gradient Boosting Classifier and Support Vector Machine Regressor algorithms stood out as having the greatest predictive power among those tested. These models can serve as valuable tools for anticipating tooth sensitivity in this patient population, facilitating better post-treatment management and control.

Integrating spatial-temporal features into prediction tasks: A novel method for identifying the potential water pollution area in large river basins

Forecasting potential water pollution areas (PWPA) is essential for effective watershed management. However, there remains a limited understanding of the spatial-temporal features that influence water quality (WQ), and advanced technical methods for WQ forecasting. This study developed an integrated framework utilizing spatial-temporal graph convolution networks (STGCN) to enhance comprehension of the spatial-temporal features influencing WQ and to develop a practical module for integrating features into the WQ prediction. The Pearson correlation method and seasonal decomposition analysis described the WQ features. Subsequently, the spatial-temporal distribution of PWPA was assessed using both the comprehensive pollution index method and Cressman space interpolation technique. Data from 403 monitoring stations was collected from the Yangtze River basin (YZR), encompassing pollutants such as COD, TP and NH₄⁺-N. The finding revealed that maximum concentrations of COD (36.4 mg/L), TP (4.078 mg/L) and NH₄⁺-N (13.58 mg/L) exceeded the standard thresholds necessitating early warnings. A significant correlation among pollutants was observed with coefficients ranging from 0.356 to 0.475 (P < 0.001), indicating their potential utility in predicting PWPA. Seasonality components exhibited strong correlations with the original WQ (correlation coefficients ranging from 0.62 to 0.89), followed by residuals (from 0.37 to 0.61) and trend components (from 0.25 to 0.53). The geographic layout of WQ monitoring stations along river lines resembled a graph network structure, suggesting that watershed WQ prediction can be classified as a spatial-temporal prediction task. The STGCN model achieved R2 values ranging from 0.607 to 0.844 for each pollutant on the test datasets, surpassing traditional models such as RNN, LSTM, and GRU in predictive accuracy. PWPA occurrences were predominantly identified in the southwestern regions as well as within the middle and lower reaches of the YZR. These results validated that the developed framework is capable of forecasting PWPA in large-scale watersheds while supporting effective watershed management.

A residual GRU method with deep cross fusion for Alzheimer’s disease progression prediction using missing variable-length time series data

Alzheimer’s disease (AD) is a neurodegenerative disorder with a long prodromal phase. Early prediction of AD progression is crucial for improving clinical diagnosis. However, missing data and variable-length of time series data make it difficult to predict the progression of AD in clinical practice. Regarding this problem, existing researches mainly employed LSTM (Long Short-Term Memory) networks to impute missing data and make predictions, but those methods only encoded fixed-length time series data and suffered from the error accumulation, resulting in relatively unsatisfactory prediction results. To address this issue, this paper develops a novel method, named residual sharing GRU (Gated Recurrent Unit) with enhanced deep cross fusion module (RGRU-EDCF), to perform AD progression prediction. Specifically, we first design an enhanced cross deep module to impute missing data and learn complementary information. Moreover, we design a residual sharing GRU model to realize the input of variable-length time series data, thus, enhancing flexibility of model. Besides, the residual structure in the residual sharing GRU model is used for reducing error accumulation. We also add a local multi-head attention mechanism for specific time series features learning for classification and regression prediction. Finally, we use specific time series features to realize prediction of classification and regression tasks. Unlike existing methods that trained the missing data imputation and prediction model separately, our method considers an end-to-end training strategy to train both enhanced cross deep module and the residual sharing GRU model to further promote the performance of AD progression prediction. The proposed method is validated using time series data from the ADNI dataset. The accuracy can reach 95.17% in the prediction task, MAE can reach 2.64 in cognitive score prediction, and 4.04 in MRI biomarkers prediction, it is competitive and superior over other state-of-the-art methods. We also validate our proposed method on MIMIC-III dataset.

Vision-Based Prediction of Flashover Using Transformers and Convolutional Long Short-Term Memory Model

The prediction of fire growth is crucial for effective firefighting and rescue operations. Recent advancements in vision-based techniques using RGB vision and infrared (IR) thermal imaging data, coupled with artificial intelligence and deep learning techniques, have shown promising solutions to be applied in the detection of fire and the prediction of its behavior. This study introduces the use of Convolutional Long Short-term Memory (ConvLSTM) network models for predicting room fire growth by analyzing spatiotemporal IR thermal imaging data acquired from full-scale room fire tests. Our findings revealed that SwinLSTM, an enhanced version of ConvLSTM combined with transformers (a deep learning architecture based on a new mechanism called multi-head attention) for computer vision purposes, can be used for the prediction of room fire flashover occurrence. Notably, transformer-based ConvLSTM deep learning models, such as SwinLSTM, demonstrate superior prediction capability, which suggests a new vision-based smart solution for future fire growth prediction tasks. The main focus of this work is to perform a feasibility study on the use of a pure vision-based deep learning model for analysis of future video data to anticipate behavior of fire growth in room fire incidents.

Multimodal market information fusion for stock price trend prediction in the pharmaceutical sector

With the evolution of China's market economy, the securities market is increasingly anchoring a pivotal role in the nation's economic landscape. Consequently, stock trend forecasting has garnered heightened attention among scholars and practitioners. This research pioneers the use of multimodal information to predict stock market fluctuations. Based on our experimental results, LSTM + Transformer performs better in handling multimodal data for stock movement prediction tasks regarding accuracy, F1-score, precision, and recall. Additionally, we employed the Granger causality test and Impulse response test to investigate the causal relationships between sentiment and stock trends, as well as the interplay between COVID-related indicators and stock trajectories. We identified discernible causal links between sentiments, COVID indicators, and stock trends for select pharmaceutical stocks. Our findings can provide valuable guidance for investors and market regulators, especially within the pharmaceutical industry. Understanding investor sentiment and the impact of the pandemic on severity can assist in effective stock commentary management and improve investment strategies.

Instant prediction of scientific paper cited potential based on semantic and metadata features: Taking artificial intelligence field as an example.

With the continuous increase in the number of academic researchers, the volume of scientific papers is also increasing rapidly. The challenge of identifying papers with greater potential academic impact from this large pool has received increasing attention. The citation frequency of a paper is often used as an objective indicator to gauge the academic influence of the paper. The task of citation frequency prediction based on historical citation data in previous studies can achieve high accuracy. However, it can only be executed after the paper has been published for a period. The delay is not conducive to timely discovery of papers with high citation frequency. In this paper, we propose a novel method for predicting cited potential of a paper based on the metadata and semantic information, which can predict the cited potential of academic paper instantly once it has been published. Specifically, the semantic information, such as abstract, semantic span and semantic inflection, is extracted to enhance the ability of the prediction model based on machine learning. To prove the effectiveness and rationality of cited potential prediction model, we conduct two experiments to validate the model and find the most effective combination of input information. The empirical experiments show that the prediction accuracy of our proposed model can reach 88% for the instant prediction of citation.

Frequency-Aware Feature Fusion for Dense Image Prediction.

Dense image prediction tasks demand features with strong category information and precise spatial boundary details at high resolution. To achieve this, modern hierarchical models often utilize feature fusion, directly adding upsampled coarse features from deep layers and high-resolution features from lower levels. In this paper, we observe rapid variations in fused feature values within objects, resulting in intra-category inconsistency due to disturbed high-frequency features. Additionally, blurred boundaries in fused features lack accurate high frequency, leading to boundary displacement. Building upon these observations, we propose Frequency-Aware Feature Fusion (FreqFusion), integrating an Adaptive Low-Pass Filter (ALPF) generator, an offset generator, and an Adaptive High-Pass Filter (AHPF) generator. The ALPF generator predicts spatially-variant low-pass filters to attenuate high-frequency components within objects, reducing intra-class inconsistency during upsampling. The offset generator refines large inconsistent features and thin boundaries by replacing inconsistent features with more consistent ones through resampling, while the AHPF generator enhances high-frequency detailed boundary information lost during downsampling. Comprehensive visualization and quantitative analysis demonstrate that FreqFusion effectively improves feature consistency and sharpens object boundaries. Extensive experiments across various dense prediction tasks confirm its effectiveness.

Task and Timing Effects in Argument Role Sensitivity: Evidence From Production, EEG, and Computational Modeling.

Comprehenders generate expectations about upcoming lexical items in language processing using various types of contextual information. However, a number of studies have shown that argument roles do not impact neural and behavioral prediction measures. Despite these robust findings, some prior studies have suggested that lexical prediction might be sensitive to argument roles in production tasks such as the cloze task or in comprehension tasks when additional time is available for prediction. This study demonstrates that both the task and additional time for prediction independently influence lexical prediction using argument roles, via evidence from closely matched electroencephalogram (EEG) and speeded cloze experiments. In order to investigate the timing effect, our EEG experiment used maximally simple Japanese stimuli such as Bee-nom/acc sting, and it manipulated the time for prediction by changing the temporal interval between the context noun and the target verb without adding any further linguistic content. In order to investigate the task effect, we conducted a speeded cloze study that was matched with our EEG study both in terms of stimuli and the time available for prediction. We found that both the EEG study with additional time for prediction and the speeded cloze study with matched timing showed clear sensitivity to argument roles, while the EEG conditions with less time for prediction replicated the standard pattern of argument role insensitivity. Based on these findings, we propose that lexical prediction is initially insensitive to argument roles but a monitoring mechanism serially inhibits role-inappropriate candidates. This monitoring process operates quickly in production tasks, where it is important to quickly select a single candidate to produce, whereas it may operate more slowly in comprehension tasks, where multiple candidates can be maintained until a continuation is perceived. Computational simulations demonstrate that this mechanism can successfully explain the task and timing effects observed in our experiments.

HiMolformer: Integrating graph and sequence representations for predicting liver microsome stability with SMILES

In the initial stages of drug discovery or pre-clinical studies, understanding the metabolic stability of new molecules is crucial. Recently, research on pre-trained deep learning for molecular property prediction has been actively progressing, with various models being made open-source. However, most of these models rely on either 2D graph or 1D sequence for training, and the representation varies depending on the data format used. Consequently, combining multiple representations can broaden the scope of learning and may potentially be a manageable and most effective method to enhance performance.Therefore, we propose a novel hybrid model for predicting metabolic stability, which integrates representations from both graph-based and sequence-based models pre-trained for molecular features. This approach utilizes the combined strengths of 2D topological and 1D sequential information of molecules. HiMol, a graph-based graph neural network (GNN) model, and Molformer, a sequence-based Transformer model, were selected for integration, thus we named it HiMolformer. HiMolformer demonstrated superior performance compared to other models. We also focus on regression task for prediction with a empirical dataset from Korea Chemical Bank (KCB), comprising 3,498 molecules with mouse liver microsome (MLM) and human liver microsome (HLM) data obtained from actual metabolic reaction experiments. To the best of our knowledge, it is the first attempt to develop MLM and HLM prediction models using regression with a single SMILES input. The source code of this model is available at https://github.com/YUNSEOKWOO/HiMolformer.

Leveraging transformer models for enhanced temperature forecasting: a comparative analysis in the Beni Mellal region

The remarkable impact of transformers in artificial intelligence, exemplified by applications like GPT-3 in language processing, has sparked interest in their potential for time series analysis. This study aims to explore whether transformers, specifically temporal fusion transformers (TFT), can outperform conventional methods in this domain. The research question is whether TFT exhibits superior performance compared to conventional recurrent neural network (RNN) methods, specifically gated recurrent unit (GRU), and traditional machine learning approaches, notably autoregressive integrated moving average (ARIMA), in the context of time series analysis and temperature prediction. A comparative analysis is conducted among three models: ARIMA, GRU, and TFT. The study utilizes time series data spanning from 1984 to the end of 2022. The models’ performances are evaluated using multiple metrics: mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE), and the coefficient of determination (R2). The TFT model achieves the lowest MAE, indicating high accuracy in its predictions. It outperforms both the RNN and traditional machine learning in temperature prediction tasks. Integrating the TFT model with the FAO penman-monteith method could improve irrigation scheduling due to more accurate temperature predictions, potentially enhancing water efficiency and crop yields.

PSTFormer: A novel parallel spatial-temporal transformer for remaining useful life prediction of aeroengine

One of the significant tasks in aeroengine remaining useful life (RUL) prediction is to address both temporal dependencies and spatial dependencies in multivariate time series (MTS) monitoring data. However, it is difficult for traditional transformer-based methods simultaneously extract both temporal and spatial dependencies due to their mutual interference, limiting the further improvement of prediction performance. To address these issues, this paper proposes a novel Parallel Spatial-Temporal Transformer (PSTFormer) for aeroengine RUL prediction with multi-sensor monitoring data. First, a novel parallel spatial–temporal attention mechanism (PSTAM) is designed, which consists of a temporal attention module (TAM) and a spatial attention module (SAM), to simultaneously capture temporal and spatial dependencies from MTS data. TAM employs multiscale convolution to learn the temporal dependencies at different time scales, while SAM adopts a self-attention mechanism to learn the spatial dependencies among different sensor parameter. Parallel connection between TAM and SAM can effectively avoid the mutual interference between temporal and spatial dependencies, improving the modeling ability of complex spatiotemporal relationships. Second, a task-guided spatiotemporal feature fusion (TG-STFF) module is designed, which adaptively fuses temporal and spatial features according to downstream task. Specifically, based on the RUL prediction characteristic, TG-STFF converts spatial features into attention weights and fuses them with temporal features to extract more representative degradation features. Finally, the effectiveness of PSTFormer is validated by a series of experimental comparisons on the public C-MAPSS dataset. Compared with SOTA methods, PSTFormer exhibits more outstanding prediction performance, and it can effectively address the aforementioned challenges in RUL prediction tasks. Therfore, the development of PSTFormer provides an innovative and effective method for aeroengine RUL prediction, significantly enhancing the efficiency and safety of aeroengine maintenance and operation.

Channel and Spatial Enhancement Network for human parsing

The dominant backbones of neural networks for scene parsing consist of multiple stages, where feature maps in different stages often contain varying levels of spatial and semantic information. High-level features convey more semantics and fewer spatial details, while low-level features possess fewer semantics and more spatial details. Consequently, there are semantic-spatial gaps among features at different levels, particularly in human parsing tasks. Many existing approaches directly upsample multi-stage features and aggregate them through addition or concatenation, without addressing the semantic-spatial gaps present among these features. This inevitably leads to spatial misalignment, semantic mismatch, and ultimately misclassification in parsing, especially for human parsing that demands more semantic information and more fine details of feature maps for the reason of intricate textures, diverse clothing styles, and heavy scale variability across different human parts. In this paper, we effectively alleviate the long-standing challenge of addressing semantic-spatial gaps between features from different stages by innovatively utilizing the subtraction and addition operations to recognize the semantic and spatial differences and compensate for them. Based on these principles, we propose the Channel and Spatial Enhancement Network (CSENet) for parsing, offering a straightforward and intuitive solution for addressing semantic-spatial gaps via injecting high-semantic information to lower-stage features and vice versa, introducing fine details to higher-stage features. Extensive experiments on three dense prediction tasks have demonstrated the efficacy of our method. Specifically, our method achieves the best performance on the LIP and CIHP datasets and we also verify the generality of our method on the ADE20K dataset.