The prevalence of long-tailed distributions in real-world data often results in classification models favoring the dominant classes,neglecting the less frequent *** approaches address the issues in long-tailed image c...
详细信息
The prevalence of long-tailed distributions in real-world data often results in classification models favoring the dominant classes,neglecting the less frequent *** approaches address the issues in long-tailed image classification by rebalancing data,optimizing weights,and augmenting ***,these methods often struggle to balance the performance between dominant and minority classes because of inadequate representation learning of the *** address these problems,we introduce descriptional words into images as cross-modal privileged information and propose a cross-modal enhanced method for long-tailed image classification,referred to as *** improves the learning of intraclass similarity of tail-class representations by cross-modal alignment and captures the difference between the head and tail classes in semantic space by cross-modal *** fusing the above information,CMLTNet achieved an overall performance that was better than those of benchmark long-tailed and cross-modal learning methods on the long-tailed cross-modal datasets,NUS-WIDE and *** effectiveness of the proposed modules was further studied through ablation *** a case study of feature distribution,the proposed model was better in learning representations of tail classes,and in the experiments on model attention,CMLTNet has the potential to help learn some rare concepts in the tail class through mapping to the semantic space.
Local search has been widely applied to solve the well-known (weighted) partial MaxSAT problem, significantly influencing many real-world applications. The main difficulty to overcome when designing a local search alg...
详细信息
Parameter-efficient finetuning (PEFT) is a key technique for adapting large language models (LLMs) to downstream tasks. In this paper, we study leveraging knowledge graph embeddings to improve the effectiveness of PEF...
详细信息
The rapid development of single-cell RNA sequencing (scRNA-seq) technology has enabled researchers to explore gene expression differences at the level of individual cells, revealing more refined cell types and states....
详细信息
Machine learning engineering is an important technology that has attracted the attention of academia and industry in the past two years. For AI to become a productivity of enterprises, it must be engineered to solve t...
详细信息
Colon cancer is a type of cancer caused by polyps that become malignant within the colon or rectum. Dealing with colon cancer effectively requires the diagnosis of the cancer at an early stage, which is of vital impor...
详细信息
High-mobility semiconductor nanotubes have demonstrated great potential for applications in high-speed transistors,single-charge detection,and memory *** we systematically investigated the electronic properties of sin...
详细信息
High-mobility semiconductor nanotubes have demonstrated great potential for applications in high-speed transistors,single-charge detection,and memory *** we systematically investigated the electronic properties of single-walled boron antimonide(BSb)nanotubes using first-principles *** observed that rolling the hexagonal boron antimonide monolayer into armchair(ANT)and zigzag(ZNT)nanotubes induces compression and wrinkling effects,significantly modifying the band structures and carrier mobilities through band folding andπ^(*)-σ^(*)*** the chiral index increases,the band gap and carrier mobility of ANTs decrease monotonically,where electron mobility consistently exceeds hole *** contrast,ZNTs exhibit a more complex trend:the band gap first increases and then decreases,and the carrier mobility displays oscillatory *** particular,both ANTs and ZNTs could exhibit significantly higher carrier mobilities compared to hexagonal monolayer and zinc-blende BSb,reaching 10^(-3)-10^(-7) cm^(-2)·V^(-1)·s^(-1).Our findings highlight strong curvature-induced modifications in the electronic properties of single-walled BSb nanotubes,demonstrating the latter as a promising candidate for high-performance electronic devices.
Sharding is a promising technique to tackle the critical weakness of scalability in blockchain-based unmanned aerial vehicle(UAV)search and rescue(SAR)*** breaking up the blockchain network into smaller partitions cal...
详细信息
Sharding is a promising technique to tackle the critical weakness of scalability in blockchain-based unmanned aerial vehicle(UAV)search and rescue(SAR)*** breaking up the blockchain network into smaller partitions called shards that run independently and in parallel,shardingbased UAV systems can support a large number of search and rescue UAVs with improved scalability,thereby enhancing the rescue ***,the lack of adaptability and interoperability still hinder the application of sharded blockchain in UAV SAR *** refers to making adjustments to the blockchain towards real-time surrounding situations,while interoperability refers to making cross-shard interactions at the mission *** address the above challenges,we propose a blockchain UAV system for SAR missions based on dynamic sharding *** from the benefits in scalability brought by sharding,our system improves adaptability by dynamically creating configurable and mission-exclusive shards,and improves interoperability by supporting calls between smart contracts that are deployed on different *** implement a prototype of our system based on Quorum,give an analysis of the improved adaptability and interoperability,and conduct experiments to evaluate the *** results show our system can achieve the above goals and overcome the weakness of blockchain-based UAV systems in SAR scenarios.
Single-cell sequencing techniques are often impacted by technical noise, leading to the generation of very sparse expression matrices. This technical noise is referred to as dropouts and poses as a major challenge for...
详细信息
With the rapid development of mobile technology and smart devices,crowdsensing has shown its large potential to collect massive *** the limitation of calculation power,edge computing is introduced to release unnecessa...
详细信息
With the rapid development of mobile technology and smart devices,crowdsensing has shown its large potential to collect massive *** the limitation of calculation power,edge computing is introduced to release unnecessary data *** edge-computing-enabled crowdsensing,massive data is required to be preliminary processed by edge computing devices(ECDs).Compared with the traditional central platform,these ECDs are limited by their own capability so they may only obtain part of relative factors and they can’t process data *** involved in one task are required to cooperate to process the task *** privacy of participants is important in crowdsensing,so blockchain is used due to its decentralization and *** crowdsensing tasks,it is usually difficult to obtain the assessment criteria in advance so reinforcement learning is *** mentioned before,ECDs can’t process task data comprehensively and they are required to cooperate quality ***,a blockchain-based framework for data quality in edge-computing-enabled crowdsensing(BFEC)is proposed in this ***(Delegated Proof of Reputation),which is proposed in our previous work,is improved to be suitable in ***,the final result is calculated without revealing the privacy of *** on the open datasets Adult,Blog,and Wine Quality show that our new framework outperforms existing methods in executing sensing tasks.
暂无评论