In this paper, we concern the uplink of a massive single-input multiple-output enabled ultra-reliable low-latency communication system, in which a single-antenna transmitter aims to timely and reliably send data to a receiver equipped with a large number of antennas over Rayleigh fading channels. For such a scenario, to eliminate the considerable overhead caused by channel estimation, we adopt a noncoherent maximum-likelihood (ML) receiver, which is known to be optimal in terms of average symbol-error rate for equiprobable discrete input signals. We propose a two-dimensional noncoherent constellation design framework to enhance the reliability of the considered system. Specifically, our design principle is to maximize the minimum Kullback-Leibler divergence between the conditional distributions induced by different transmitted signals under average power constraint for any given transmission rate. The resulting optimization problem is shown to be a challenging mixed discrete-continuous problem. We manage to solve the problem by deliberately designing optimal bit allocation and optimal constellation structure as a function of signal-to-noise ratios. We then unveil that the proposed constellation can facilitate efficient ML detection with low computational complexity. Finally, simulation results illustrate that the proposed scheme has a superior error performance than conventional training-based schemes and existing energy detection designs.

Semantic segmentation is a fundamental task in multimedia processing, which can be used for analyzing, understanding, editing contents of images and videos, among others. To accelerate the analysis of multimedia data, existing segmentation researches tend to extract semantic information by progressively reducing the spatial resolutions of feature maps. However, this approach introduces a misalignment problem when restoring the resolution of high-level feature maps. In this paper, we design a Semantic Refinement Module (SRM) to address this issue within the segmentation network. Specifically, SRM is designed to learn a transformation offset for each pixel in the upsampled feature maps, guided by high-resolution feature maps and neighboring offsets. By applying these offsets to the upsampled feature maps, SRM enhances the semantic representation of the segmentation network, particularly for pixels around object boundaries. Furthermore, a Contextual Refinement Module (CRM) is presented to capture global context information across both spatial and channel dimensions. To balance dimensions between channel and space, we aggregate the semantic maps from all four stages of the backbone to enrich channel context information. The efficacy of these proposed modules is validated on three widely used datasets-Cityscapes, Bdd100K, and ADE20K-demonstrating superior performance compared to state-of-the-art methods. Additionally, this paper extends these modules to a lightweight segmentation network, achieving an mIoU of 82.5% on the Cityscapes validation set with only 137.9 GFLOPs.

Semi-supervised learning (SSL) utilizes limited labeled data and plenty of unlabeled data, and it has attracted attentions for its improved learning performance. However, recent studies have indicated that using unlabeled data, in some cases, could deteriorate the performance. Therefore, there’s an imminent need to develop safe semi-supervised learning methods to determine whether SSL should be applied for a given scenario. This paper proposes a safe version of multi-class graph-based semi-supervised support vector machine (SVM). At first, in order to eliminate the impact of bad label assignments, a criterion based on the cost function of semi-supervised SVM (S3VM) is introduced to evaluate the predicted label assignments. Then, m candidate optimal label assignments are picked up by the criterion. After that, a multi-class safe strategy is designed to generate the final label assignment whose performance is never worse than that of the methods using only labeled samples. Experimental results on several benchmark datasets validate the effectiveness of the proposed technique.

Local feature descriptor learning aims to represent distinctive images or patches with the same local features, where their representation is invariant under different types of deformation. Recent studies have demonstrated that descriptor learning based on Convolutional Neural Network (CNN) is able to improve the matching performance significantly. However, they tend to ignore the importance of sample selection during the training process, leading to unstable quality of descriptors and learning efficiency. In this paper, a dual hard batch construction method is proposed to sample the hard matching and non-matching examples for training, improving the performance of the descriptor learning on different tasks. To construct the dual hard training batches, the matching examples with the minimum similarity are selected as the hard positive pairs. For each positive pair, the most similar non-matching example is then sampled from the generated hard positive pairs in the same batch as the corresponding negative. By sampling the hard positive pairs and the corresponding hard negatives, the hard batches are produced to force the CNN model to learn the descriptors with more efforts. In addition, based on the above dual hard batch construction, an ℓ22 triplet loss function is built for optimizing the training model. Specifically, we analyze the superiority of the ℓ22 loss function when dealing with hard examples, and also demonstrate it in the experiments. With the benefits of the proposed sampling strategy and the ℓ22 triplet loss function, our method achieves better performance compared to state-of-the-art on the reference benchmarks for different matching tasks.

In recent years, multitrack music generation has garnered significant attention in both academic and industrial spheres for its versatile utilization of various instruments in collaborative settings. The primary challenge lies in achieving a harmonious balance within individual tracks and fostering effective collaboration across multiple tracks. To address this issue, this article introduces a pioneering hybrid learning encoder architecture. Each music track's encoder is implemented as an independent transformer architecture, preserving self-attention mechanisms within a single track and interattention mechanisms between different tracks. The resulting features are then seamlessly integrated into the decoder through concatenation. Of particular significance, previous multitrack music generation efforts have predominantly operated under unconditional settings, yielding music that lacks practical value due to noncompliance with established music theory principles. Recognizing this limitation, the article proposes a novel approach to multitrack music generation guided by music theory rules. Employing reinforcement learning techniques, the decoder-generated music serves as the initial state. Positive feedback is provided when the generated music adheres to music theory rules; conversely, negative feedback is applied to compel the multitrack music to align with widely accepted music theory principles. Finally, comprehensive simulation validation is conducted on both the publicly available LMD dataset and the self-constructed MUT dataset. The plethora of experimental results overwhelmingly corroborates the efficacy of the proposed methodology.