You-Only-Look-Once (YOLO) represents the state-of-the-art in object detection models. With the emergence of various applications utilizing small domain-specific datasets and limited computing resources for extensive model training and deployment, there is an increasing demand for customized lightweight YOLO architectures. In this paper, we propose a general NeuroEvolution-based method, termed YOLO-Light, designed to automatically create lightweight variants of YOLO architectures tailored to object detection tasks across diverse scenarios. For a given task, YOLO-Light first initializes a population of minimal YOLO architectures and subsequently evolves these models within a novel parallel-chain evolutionary space. This process employs a diversity-protecting evolutionary search strategy until some architectures meet the expected performance standards. During evolution, YOLO-Light incorporates a dynamic evolution regulation mechanism to adjust the evolutionary configuration, thereby enhancing efficiency based on the current evolutionary state. We applied YOLO-Light to generate lightweight YOLOv5, YOLOv8, and YOLOv10 architectures for object detection on the Roboflow 100 small dataset collection, which comprises 100 diverse datasets spanning 7 distinct imagery domains, with a total of 224,714 images and 829 classes. Our experiments focused on 20 datasets ranging from 105 to 8,992 images and 1 to 53 classes. The experimental results show that YOLO-Light reduced the number of parameters by 54–95%, while maintaining or improving mean Average Precision (mAP) compared to standard YOLO architectures. These results demonstrate the effectiveness of YOLO-Light in generating lightweight, task-specific YOLO architectures for resource-constrained object detection tasks. The code repository of YOLO-Light is available on GitHub at https://github.com/BruceShine/YOLO-Light.

Analyzing gigapixel Whole Slide Images (WSIs) is challenging due to the complex pathological tissue environment and the absence of target-driven domain knowledge. Previous methods incorporated pathological priors to mitigate this issue but relied on additional inference steps and specialized workflows, restricting scalability and the model’s capacity to identify novel outcome-related factors. To address these challenges, we propose a plug-and-play Pathology-Aware Mixture-of-Experts (PAMoE) module, which based on mixture of experts to learn pathology-related knowledge and extract useful information. We train the experts to become ‘specialists’ in specific intratumoral tissues by learning to route each tissue to its mapped expert. In addition, to reduce the impact of irrelevant content on the model, we introduce a new routing rule that discards patches in which none of the experts express interest, which helps the model better capture the relationships between relevant patches. Through a comprehensive evaluation of PAMoE on survival task, we demonstrate that 1) Our module enhances the performance of baseline models in most cases, and 2) The sparse expert processing across different tissues enhances the learning of patch representations by addressing tissue heterogeneity. Source code is available at https://github.com/wjx-error/PAMoE.

Survival prediction is a significant challenge in cancer management. Tumor micro-environment is a highly sophisticated ecosystem consisting of cancer cells, immune cells, endothelial cells, fibroblasts, nerves and extracellular matrix. The intratumor heterogeneity and the interaction across multiple tissue types profoundly impacts the prognosis. However, current methods often neglect the fact that the contribution to prognosis differs with tissue types. In this paper, we propose ProtoSurv, a novel heterogeneous graph model for WSI survival prediction. The learning process of ProtoSurv is not only driven by data but also incorporates pathological domain knowledge, including the awareness of tissue heterogeneity, the emphasis on prior knowledge of prognostic-related tissues, and the depiction of spatial interaction across multiple tissues. We validate ProtoSurv across five different cancer types from TCGA (i.e., BRCA, LGG, LUAD, COAD and PAAD), and demonstrate the superiority of our method over the state-of-the-art methods.

Lane detection is a special task in autonomous driving. Its most prominent inherent feature is to learn the imagination of severely occluded objects. Traditional CNN-based networks learning the imagination tend to perform poorly. In this work, we propose a novel architecture, called Cycle_accumulation-Transformer (CaT), which is the first structure to handle the lane detection by fusing CNN and Transformer. In particular, Cycle_accumulation structure and Transformer structure complement each other, and they adopt the four-direction cyclic accumulation process of “up to down”, “down to up”, “left to right” and “right to left” in the convolutional mode and the self-attention mechanism of “QKV” to fuse global information respectively. Our method is based on pixel-level semantic segmentation with high detection accuracy while meeting real-time requirements. Moreover, our proposed method achieves state-of-the-art results on the Tusimple and also achieves competitive results on the CULane.

Precise delineation of hippocampus subfields is crucial for the identification and management of various neurological and psychiatric disorders. However, segmenting these subfields automatically in routine 3T MRI is challenging due to their complex morphology and small size, as well as the limited signal contrast and resolution of the 3T images. This research proposes Syn_SegNet, an end-to-end, multitask joint deep neural network that leverages ultrahigh-field 7T MRI synthesis to improve hippocampal subfield segmentation in 3T MRI. Our approach involves two key components. First, we employ a modified Pix2PixGAN as the synthesis model, incorporating self-attention modules, image and feature matching loss, and ROI loss to generate high-quality 7T-like MRI around the hippocampal region. Second, we utilize a variant of 3D-U-Net with multiscale deep supervision as the segmentation subnetwork, incorporating an anatomic weighted cross-entropy loss that capitalizes on prior anatomical knowledge. We evaluate our method on hippocampal subfield segmentation in paired 3T MRI and 7T MRI with seven different anatomical structures. The experimental findings demonstrate that Syn_SegNet's segmentation performance benefits from integrating synthetic 7T data in an online manner and is superior to competing methods. Furthermore, we assess the generalizability of the proposed approach using a publicly accessible 3T MRI dataset. The developed method would be an efficient tool for segmenting hippocampal subfields in routine clinical 3T MRI.

Polar codes are the first provable capacity-achieving channel codes. Despite the splendid performance of long Polar codes, short Polar codes have relatively poor performance compared with other modern channel coding schemes (e.g., Turbo codes and LDPC). In this paper, we explore some practical methods to improve the performance of Polar codes with short to moderate codeword lengths. First, we use Reed Solomon (RS) codes as outer codes. With a specific interleaving strategy, we can concatenate multiple RS codes with one frame of Polar codes. Combining a strategy of allocating unequal RS code rates with the concatenation, different levels of protection are assigned based on the error pattern of successive cancellation list (SCL) decoders. Thus, the finite length performance will certainly be enhanced for this encoding scheme. Meanwhile, the memory size that the original SCL decoding procedure requires is reduced, and the increment of overall decoding complexity is small. Additionally, we propose an intra-frame interleaver to further enhance the performance by dispersing errors. Finally, we designed a list decoding scheme for the proposed multi-RS concatenated Polar codes. Depending on the soft information generated by an SCL decoder, we calculated the reliability of each RS symbol and conducted soft RS decoding. So, the overall performance was enhanced under this joint decoding strategy. Simulation results indicate that the bit error rate (BER) performance of short Polar codes can be well improved.

Ideal Hall sensors installation interval is 120° electrical angle, however, the inevitable mechanical installation error may cause Hall signals delay or advance, aggravating seriously electromagnetic torque ripple. This paper presents a novel compensation method for commutation torque ripple reduction of Brushless DC motor (BLDC motor) with misaligned Hall sensors. First of all, by comparing phase current stable value with outgoing phase current detected at rising edge of the corresponding Hall signal, Hall sensors delayed or advanced installation is determined. And then, Hall sensors installation error angle can be obtained by deriving the relationship between three phase currents and error angle, which can be compensated by the optimal delayed angle for minimizing commutation torque ripple. At last, the simulation results verify the effectiveness of the theories and the feasibility of the proposed method.

The conventional pulse width modulation (PWM) modes for braking operation in Brushless DC motor may cause diode freewheeling in inactive phase during normal conduction period which will aggravate torque ripple. In this paper, a novel braking PWM modulation mode named PWM_OFF_PWM is proposed to eliminate the diode freewheeling. The performance of the proposed PWM_OFF_PWM mode on diode freewheeling and commutation torque ripple are analyzed. During normal conduction period, the proposed PWM_OFF_PWM mode is confirmed to give no contribution to diode freewheeling; during commutation period, the speed range where commutation torque ripple can be suppressed is derived. At last, simulation results verify the feasibility of the proposed braking PWM_OFF_PWM mode and the correctness of the theoretical analysis.

As the first ever provably capacity achieving codes, Polar codes have drawn a wide range of research interests in recent years. It is well known that short/finite-length Polar codes have relatively not so good bit error rate (BER) performance as the state-of-the-art channel codes (e.g. Turbo codes, LDPC). One commonly used way to improve the performance of short Polar codes is to concatenate the Polar codes with outer codes, but the amount of improvement is largely constrained by the performance of the outer codes with short codeword length. Motivated by this, in this work, we propose to use the newly invented Spinal codes, which has high performance with short code length, as the outer codes. Specifically, the designed codes, named as Spinal-Polar, is implemented through an interleaved concatenation scheme. In addition, we propose a joint iterative decoding algorithm for SpinalPolar, and the decoding complexity is analyzed theoretically. Extensive simulations are carried out, and results show that the proposed concatenation scheme can significantly improve the BER performance of short Polar codes.

By combining channel coding and virtual MIMO transmission, coded cooperation could achieve coding gain and diversity gain simultaneously, making it a good candidate for the key technologies enabling ultra-high speed 5G communications. As the first ever provably capacity achieving codes, Polar codes naturally sticks out to be one of the most competitive coding technologies for coded cooperation. In this paper, we aim to propose methods that can fully explore the performance potential of Polar coded cooperation under fading channels. Specifically, Polar coded cooperation by adopting the Plotkin construction for sub-codeword generation is used as the basic method. Then, three codeword shaping methods are proposed to improve the performance of the basic method. The first one is to introduce an interleaver at the receiver terminal to help combat the burst errors. On this basis, the idea of information-refreezing is used to improve the sub-codeword decoding performance on the interuser channels, which in turn increases the cooperation probability. And lastly, the codeword generation scheme is extended from non-systematic Polar codes to systematic Polar codes so that a systematic coding gain is further achieved. The proposed three shaping methods can be used either singly or superimposedly. Simulation results show that under slow fading channels, the system performance in terms of bit error rate can be significantly improved over that of existing Polar coded cooperation method.

This paper presents a novel method based on delaying Hall signal for reducing torque ripple of brushless DC motor. During the delayed interval, the back electromotive force (back-EMF) amplitude of outgoing phase declines and the phase current rises to stabilize phase voltage. The electromagnetic torque can remain steady by regulating the duty ratio. During the commutation interval, the time of outgoing phase current decays to zero is equivalent to the time of incoming phase current reaches stable value, which can derive the delayed time to minimize the commutation torque ripple. The method used in this paper with no need for any additional electrical devices, can obtain the delayed time under knowing the motor parameters. At last, the results of the simulation in MATLAB verify the correctness of the theories and the effectiveness of the proposed approach.