Transformer-based models have demonstrated great promises in single image super-resolution (SISR), but our investigations find significant redundancy in terms of high mutual information across the attention maps, which is associated with reduced efficiency and degraded performance of SOTA models. To address the problem, here we propose a low redundancy attention network (LRAN). First, to mitigate the redundancy among heads, we introduce in the self-attention computation a multi-element mechanism, which allows for the incorporation of various types of self-attention, thus increasing inter-head diversity. Second, to address the redundancy among blocks, we propose the encapsulated architecture, in which enhanced local perception unit and gated multi-layer perceptron are designed to capture local information. Specifically, this architecture incorporates a single self-attention layer between several MLP layers. Subsequently, the proposed gated multi-layer perceptron significantly enhances the SR quality. Extensive experiments demonstrate that LRAN outperforms SOTA models in the task of lightweight SR, achieving a better trade-off between quality and speed. For instance, the proposed LRAN-light surpasses SwinIR-light by 0.32dB PSNR in $\times 4$ SR on Urban100, while running $\times 4$ faster.

In the field of blind image quality assessment, accurately capturing localized distortions and structural inconsistencies within images remains a significant challenge. To tackle this issue, we propose BlockIQA, a novel framework that enhances local sensitivity through deep block analysis. BlockIQA divides images into non-overlapping blocks and employs a multi-branch architecture that integrates ResNet50, the Feature Pyramid Network, and an Auxiliary Feature Extraction Layer. The primary innovations of this paper include: (1) Segmenting images into smaller blocks for detailed analysis and employing a Gaussian similarity model that dynamically adapts to variations in feature dimensions and directional consistency. This approach enables more precise characterization of local image features. (2) Achieving a balance between global semantic information and localized distortion patterns through multiscale feature fusion using feature pyramid networks and auxiliary feature extraction layer. This ensures that while capturing overall image semantics, no fine local distortion information is overlooked. Experimental results demonstrate that BlockIQA performs well across datasets with various types of distortions and exhibits strong generalizability across different databases. In summary, BlockIQA pioneers a new deep learning architectural paradigm for blind image quality assessment. Its design philosophy of enhancing local sensitivity through deep block analysis provides valuable new ideas and methods for research and practice in this domain.

Time series forecasting is crucial in areas like smart grids, traffic flow management, and financial analysis, especially for Long-sequence Time Series Forecasting (LTSF) tasks. We propose WaveSparseTSF, a lightweight LTSF model designed to handle complex temporal dependencies while minimizing computational costs. It integrates wavelet transform and cross-period sparse forecasting, which separate the data into low and high frequency components and focus on periodic trends to reduce complexity. By processing raw sequences through wavelet decomposition and downsampling, WaveSparseTSF effectively captures both global trends and local fluctuations. Despite using only around 1k parameters, it achieves competitive or superior results compared to state-of-the-art models and demonstrates strong generalization capabilities. As a result, WaveSparseTSF is particularly well-suited for environments with limited resources, small datasets, or low-quality data.

Hand-eye calibration is a crucial step in the implementation of vision-based robotic arm systems. However, existing calibration methods struggle to adapt to scenarios where the robotic arm or vision system frequently changes position. To address this challenge, this paper proposes a novel calibration method based on an exploration-optimized Twin Delayed DDPG (TD3) algorithm enhanced with kernel density estimation and decaying noise, referred to as E-TD5. The proposed approach not only incorporates the E-TD5 algorithm to improve the exploration and exploitation capabilities of the TD3 reinforcement learning framework but also introduces an adaptive target-perception enhancement system to handle frequent variations in hand-eye positioning. Experimental results validate the effectiveness of the proposed method, its robustness to changes in hand-eye positions, and the significant advantages of the E-TD5 algorithm compared to the standard TD3 approach.

Transformer-based models have demonstrated impressive performance in image super-resolution (SR), but they come with a high computational overhead. In this paper, we present a low redundant attention network (LRAN) for efficient image SR. We observe that there is significant similarity in attention maps across heads and blocks, leading to computational redundancy. First, to mitigate the redundancy in attention maps among heads, we introduce a multi-element mechanism in the self-attention computation. This mechanism allows for the incorporation of various types of self-attention, thus increasing inter-head diversity. Second, to address this redundancy in attention maps among blocks, we propose the hamburger architecture, which introduces enhanced local perception units to capture local information. Moreover, this architecture incorporates a single self-attention layer between several efficient MLP layers. Extensive experiments demonstrate that LRAN outperforms the latest models in lightweight SR, achieving a better trade-off between SR quality and latency. For instance, LRAN surpasses SwinIR-light by 0.25dB PSNR in ×4 SR on Urban100, while running ×5 faster.

The Internet-of-Things (IoT) technology is becoming increasingly pivotal in the financial services sector, with a growing number of algorithms being employed in high-frequency trading. High-frequency prediction in financial time series prediction presents a promising avenue of research. From convolutional neural networks to recurrent neural networks, deep learning have demonstrated exceptional capabilities in capturing the nonlinear characteristics of stock markets, thereby achieving high performance in stock index prediction. In this paper, we employ ODE-LSTM model for high-frequency price forecasting, predicting stock price data across various time scales, including 1-minute, 5-minutes, and 30-minutes frequencies. This approach introduces a novel concept, wherein the LSTM (Long Short-Term Memory) model is integrated with Neural ODE (Ordinary Differential Equations) to manage the hidden state and augment model interpretability. Over the course of 7 months, we achieved a 41.79% excess return on a simulated trading platform, with a daily average excess return of 0.30%, showcasing the commendable performance of our model and strategy.