Compound fault diagnosis in multi-joint industrial robots is a critical yet underexplored problem in industrial internet of things, where the simultaneous degradation of multiple joints poses a severe challenge for reliable operation. Unlike conventional methods limited to single-fault scenarios, this paper addresses the compositional generalization challenge—requiring models trained only on simple faults to accurately recognize unseen higher-order fault compositions. To this end, we propose StateMix Network (SMNet), a multi-stage architecture that preserves atomic joint-level representations before compositional diagnosis. Specifically, a Single-Joint Feature Extraction (SJFE) backbone extracts clean joint-private features, which are then fused by an Attention-Guided Dilated Fusion (AGDF) neck employing parallel Cascaded Dilated Convolution Blocks (CDCBs) bracketed by a dual-path attention mechanism for scale- and context-aware integration. Finally, a Mamba-based sequence mixer models long-range cross-joint dependencies to capture global fault dynamics. Extensive experiments on in-situ vibration data from a single six-joint industrial robot platform, under a strict train-on-simple/evaluate-on-complex protocol, demonstrate that SMNet consistently outperforms representative baselines in macro-Precision, Recall, and F1-score, particularly on unseen triple- and quadruple-joint compositions. Ablation and sensitivity analyses further validate the effectiveness of each module. This work presents a diagnostic approach that effectively generalizes from simple to complex fault scenarios in industrial robots.

Single-cell RNA-seq data has become a critical source in revealing cellular activities. However, the developing probing techniques and the fatal damages to the detected cells incur various kinds of noise, e.g. the batch effect and the absence of cellular correspondence between experimental groups. Therefore, many single-cell tasks are better modeled as generative rather than discriminative tasks, since, instead of the exact cell-wise ground truth, only the distribution of cellular profiles under a certain condition is measurable. Considering the highly nonlinear and complex associations between gene expressions, we developed scVDM, a latent diffusion model integrated with a transformer-based conditional denoiser to learn three different generative tasks in single-cell data, including conditional data generation, batch effect correction and drug perturbation prediction. The high dimensional transcriptomic data are firstly projected to the latent space through a conditional VAE and then the complicated relationships between latent dimensions are deeply exploited through self-attentions to generate realistic diffusion noise. Based on the evaluation of five real-world datasets, our method demonstrates outstanding performance through comprehensive experimental results in all generative tasks.

Two-photon lithography (TPL) is an advanced technique used for additive manufacturing. How to effectively inspect the part quality is one of the challenges of TPL before large-scale industrial application. To produce cured part, the light dosage parameter is limited during the fabrication process, and the limit varies from different application scenarios. By automatic recognition of part quality, engineers can efficiently find light dosage limits and monitor the fabrication process. This paper introduces a visual monitoring-based video Transformer with three-dimensional (3D) shifted window multi-head self-attention for automatically detecting part quality in four typical real scenarios. This framework introduces a multi-head self-attention mechanism to capture global features, thereby integrating spatial and sequential information for part quality recognition. The 3D shifted window mechanism is also applied to introduce the locality similar to convolution and reduce computational complexity. In addition, hierarchical representation is introduced to Transformer architecture, which helps to model high-level information from low-level features. The dataset with four scenarios, which are different in write pattern and photoresist, is used to evaluate the feasibility of the industrialization of this framework. The results show that the proposed method has better performance than the traditional deep learning model in the detection of part quality.

Traditional carbon footprint measurement approaches fail to address the unique requirements of green music production, where conventional lifecycle assessment methods cannot accommodate the dual imperatives of environmental impact reduction and acoustic performance preservation. Existing carbon accounting frameworks lack the granularity to capture real-time operational variations in creative environments, while standard IoT monitoring systems ignore the acoustic signatures that indicate both energy efficiency and production quality. This paper introduces harmonic music carbon assessment and prediction (HMCAP), a novel dual-phase methodology that harmonizes environmental impact assessment with acoustic performance requirements throughout green music production supply chains. The first phase divides the music production supply chain into five distinct stages and develops corresponding measurement models enhanced with acoustic-sensitive AIoT sensor networks, enabling quantitative carbon footprint evaluation across raw materials, manufacturing, distribution, usage, and end-of-life stages. The second phase constructs a long short-term memory networks - based prediction framework with parallel processing pathways for acoustic and energy features, creating a specialized carbon emission forecasting system for green music production environments. We validated our approach using two years of operational data from a leading music equipment manufacturer and showed a 99.4% prediction accuracy. Implementation resulted in 14.6% carbon intensity reduction across manufacturing operations and 22.3% decrease in recording session emissions, with 92% of artists reporting no perceptible quality difference—confirming HMCAP's capacity to balance sustainability with creative excellence in green music production.

The ability to identify and track T-cell receptor (TCR) sequences from patient samples is becoming central to the field of cancer research and immunotherapy. Tracking genetically engineered T cells expressing TCRs that target specific tumor antigens is important to determine the persistence of these cells and quantify tumor responses. The available high-throughput method to profile TCR repertoires is generally referred to as TCR sequencing (TCR-Seq). However, the available TCR-Seq data are limited compared with RNA sequencing (RNA-Seq). In this paper, we have benchmarked the ability of RNA-Seq-based methods to profile TCR repertoires by examining 19 bulk RNA-Seq samples across 4 cancer cohorts including both T-cell-rich and T-cell-poor tissue types. We have performed a comprehensive evaluation of the existing RNA-Seq-based repertoire profiling methods using targeted TCR-Seq as the gold standard. We also highlighted scenarios under which the RNA-Seq approach is suitable and can provide comparable accuracy to the TCR-Seq approach. Our results show that RNA-Seq-based methods are able to effectively capture the clonotypes and estimate the diversity of TCR repertoires, as well as provide relative frequencies of clonotypes in T-cell-rich tissues and low-diversity repertoires. However, RNA-Seq-based TCR profiling methods have limited power in T-cell-poor tissues, especially in highly diverse repertoires of T-cell-poor tissues. The results of our benchmarking provide an additional appealing argument to incorporate RNA-Seq into the immune repertoire screening of cancer patients as it offers broader knowledge into the transcriptomic changes that exceed the limited information provided by TCR-Seq.

Building a quality service-based system (SBS) is one of the most important research topics in software engineering. Many studies investigate intelligent methods to simplify the process of building SBSs. In particular, some keyword-based SBS building methods allow service users to automatically build an SBS by only providing a few of keywords. This type of work usually constructs a directed weighted graph of a service repository. A set of minimum-weight group Steiner trees (MSTs) is extracted from the graph to represent the service functions and their relations. However, to the best of our knowledge, none of the existing keyword-based SBS building methods allow the relaxation of the function requirements for a user. A relaxed SBS may achieve a comparable functionality versus a complete SBS containing all the query functions. To fill in the above gap, we define a new problem: a bounded skyline SBS building problem, whose solution is more adaptive and less limited than the traditional keyword-based SBS building methods. To solve this problem, we propose two algorithms based on skyline query, dynamic programming, and lower bound pruning. In the experiments, we collect real-world datasets and label the nodes with keywords. We conduct a comprehensive study to demonstrate the time efficiency of our algorithms on automatically finding SBSs. We make the annotated real-world datasets and our source code open to peer researchers.

In this article, we consider the timeliness of information transmissions in a three-node industrial wireless sensor network (IWSN) in terms of Age of Information (AoI). In this network, a sensor monitors the ambient environment and transmits the sensed information to a remote monitor directly or through a relay node. In particular, we are interested in how the timeliness of the system is changed by decomposing the long-distance transmission with a relay and by enabling parallel transmissions over the two hops with a packet buffer. To this end, we derive the average AoIs of the transmissions over the direct-link, the relay-links with and without a buffer in a closed form. The obtained results show that the relay-link with a buffer outperforms the other two links, while the relay-link without a buffer outperforms the direct-link only if the relay is properly placed and the sensor–monitor distance is relatively large. On the condition that the average transmission times over the direct-link and the relay-link without a buffer are equal, we further evaluate how fast the average AoI can be reduced by using a relay or a packet buffer, as the packet rate approaches the maximum feasible rate over the links. It is shown that, although the sensor–monitor distance dominates the average AoIs of the links, the gains of using the relay and the buffer do not change much with the distance and are approximately constant.

Accurate and real-time fault diagnosis (FD) and working conditions identification (WCI) are the key to ensuring the safe operation of mechanical systems. We observe that there is a close correlation between the fault condition and the working condition in the vibration signal. Most of the intelligent FD methods only learn some features from the vibration signals and then use them to identify fault categories. They ignore the impact of working conditions on the bearing system, and such a single-task learning method cannot learn the complementary information contained in multiple related tasks. Therefore, this article is devoted to mining richer and complementary globally shared features from vibration signals to complete the FD and WCI of rolling bearings at the same time. To this end, we propose a novel multitask attention convolutional neural network (MTA-CNN) that can automatically give feature-level attention to specific tasks. The MTA-CNN consists of a global feature shared network (GFS-network) for learning globally shared features and K task-specific networks with feature-level attention module (FLA-module). This architecture allows the FLA-module to automatically learn the features of specific tasks from globally shared features, thereby sharing information among different tasks. We evaluated our method on the wheelset bearing data set and motor bearing data set. The results show that our method has a better performance than the state-of-the-art deep learning methods and strongly prove that our multitask learning mechanism can improve the results of each task.

The critical issue for fault diagnosis of wheel-set bearings in high-speed trains is to extract fault features from vibration signals. To handle high complexity, strong coupling, and low signal-to-noise ratio of the vibration signals, this article proposes a novel multibranch and multiscale convolutional neural network that can automatically learn and fuse abundant and complementary fault information from the multiple signal components and time scales of the vibration signals. The proposed method combines the conventional filtering methods and the idea of the multiscale learning, which can extend the breadth and depth of the feature learning process. Consequently, the proposed network can perform better. The experimental results on the wheelset bearing dataset demonstrate that the proposed method has better antinoise ability and load domain adaptability and can diagnose 12 fault types more accurately when compared with the five state-of-the-art networks.

Recently, deep-learning-based fault diagnosis methods have been widely studied for rolling bearings. However, these neural networks are lack of interpretability for fault diagnosis tasks. That is, how to understand and learn discriminant fault features from complex monitoring signals remains a great challenge. Considering this challenge, this article explores the use of the attention mechanism in fault diagnosis networks and designs attention module by fully considering characteristics of rolling bearing faults to enhance fault-related features and to ignore irrelevant features. Powered by the proposed attention mechanism, a multiattention one-dimensional convolutional neural network (MA1DCNN) is further proposed to diagnose wheelset bearing faults. The MA1DCNN can adaptively recalibrate features of each layer and can enhance the feature learning of fault impulses. Experimental results on the wheelset bearing dataset show that the proposed multiattention mechanism can significantly improve the discriminant feature representation, thus the MA1DCNN outperforms eight state-of-the-arts networks.