It has been recognized that the data generated by denoising diffusion probabilistic model (DDPM) improves adversarial training. After two years of rapid development in models, a question naturally arises: can better models further improve training? This paper gives an affirmative answer employing most recent which higher efficiency ($\sim 20$ sampling steps) and image quality (lower FID score) compared with DDPM. Our adversarially trained achieve state-of-the-art performance on RobustBench...

Intelligent agents stand out as a potential path toward artificial general intelligence (AGI). Thus, researchers have dedicated significant effort to diverse implementations for them. Benefiting from recent progress in large language models (LLMs), LLM-based that use universal natural an interface exhibit robust generalization capabilities across various applications -- serving autonomous general-purpose task assistants coding, social, and economic domains, offer extensive exploration...

Most of heterogeneous information network (HIN) based recommendation models are on the user and item modeling with meta-paths. However, they always model users items in isolation under each meta-path, which may lead to extraction misled. In addition, only consider structural features HINs when during exploring HINs, useful for lost irreversibly. To address these problems, we propose a HIN unified embedding recommendation, called HueRec. We assume there exist some common characteristics...

Sepsisis among the leading causes of morbidity and mortality in modern intensive care units. Accurate sepsis prediction is critical importance to save lives reduce medical costs. The rapid advancements sensing information technology facilitate effective monitoring patients' health conditions, generating a wealth data, provide an unprecedented opportunity for data-driven diagnosis sepsis. However, real-world data are often complexly structured with high level uncertainty (e.g., missing...

Consumer reviews are an important source of data used to judge and examine consumer sentiment, mining for electronic products is way help improve the design products. The research based on online cell phone e-commerce, paper constructs a sentiment dictionary in this field Sentiment Oriented Point Mutual Information (SO-PMI) algorithm, weight review word vectors. An extreme Gradient Boosting Tree (XGBoost) integrate vectors Large Language Model (LLM) construct recognition model, finally,...

The geometric designs of MEMS devices can profoundly impact their physical properties and eventual performances. However, it is challenging for researchers to rationally consider a large number possible designs, as would be very time- resource-consuming study all these cases using numerical simulation. In this paper, we report the use deep learning techniques accelerate design cycle by quickly accurately predicting numerous candidates with vastly different features. Design are represented in...

This paper reports the use of machine learning in accelerating MEMS design process. Candidate designs are represented by pixelated binary 2D images. Instead common computational tools like FEA, we trained neural network for quickly obtaining physical properties interest each candidate design. Circular disk resonators used as an example to demonstrate capability our method. After sufficient training with 9000 images, resulting can serve a high-speed, high-accuracy analyzer: it identify four...

Yuren Mao, Zekai Wang, Weiwei Liu, Xuemin Lin, Wenbin Hu. Proceedings of the 59th Annual Meeting Association for Computational Linguistics and 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

With the construction of Power Internet Things (PIoT) in full swing as well development wireless communication technology, deployment sensor networks is key to intelligent transformation power systems. This paper proposes an asymmetric double-layer network coverage scheme for substation. We conducted field measurements substation record received at different locations, which were compared with prediction results simulation model established by Winprop verify effectiveness method. Based on...

Myocardial infarction (MI), also known as heart attack, is the leading cause of death in United States. Accurate MI prediction critical importance to reduce healthcare costs and save lives. Rapid developments data infrastructure information technology provide an unprecedented opportunity for data-driven prediction. However, real-world medical are generally subject a high level uncertainty with imbalanced issue considerable missing values, which pose significant challenges reliable disease...

With the rapid development of generation model, AI-based face manipulation technology, which called DeepFakes, has become more and realistic. This means forgery can attack any target, poses a new threat to personal privacy property security. Moreover, misuse synthetic video shows potential dangers in many areas, such as identity harassment, pornography news rumors. Inspired by fact that spatial coherence temporal consistency physiological signal are destroyed generated content, we attempt...

Multi-task Learning (MTL), which involves the simultaneous learning of multiple tasks, can achieve better performance than each task independently. It has achieved great success in various applications, ranging from Computer Vision (CV) to Natural Language Processing (NLP). In MTL, losses including tasks are jointly optimized. However, it is common for these be competing. When competing, minimizing some increases others, accordingly variance (variance between task-specific loss);...

The era of big data has made vast amounts clinical readily available, particularly in the form electronic health records (EHRs), which provides unprecedented opportunities for developing data-driven diagnostic tools to enhance decision making. However, application EHRs modeling faces challenges such as irregularly spaced multi-variate time series, issues incompleteness, and imbalance. Realizing full potential hinges on development advanced analytical models. In this paper, we propose a novel...

Sea otters are in great danger nowadays while river free of worrying about predators, and they occupy some common habitats. Hence, it is meaningful to develop a Convolutional Neural Network (CNN) classifier aiming at distinguishing these two species when rescuers confused. This paper illustrates developed model MobileNetV2 with the support Block Attention Module (CBAM) attention module, well-performed recognizing accuracy, memory usage, time consumption portability. The result shows that...

Reciprocal recommendation is the core of many social websites like online recruitment and dating. Most recently, graph neural networks have been exploited by few researchers for reciprocal recommendation. However, they tend to oversimplify interactions between users, treating them as simple pairwise relationships, which overlooks multidimensional relationships among users. Additionally, these methods fail consider users' historical interaction sequences feedback behaviors, makes it...

This paper is here to develop a model that can assess the health status of higher education system in any country. Then select country and propose set policies will move from its current state target state. Five indicators for evaluation are formulated. Then, Python crawler used capture data different countries under these five Internet. K-means++ clustering method classify into three classes. The significance specific results classification shown 4.2.1 Table3. according types countries,...

This paper is here to develop a model that can assess the health status of higher education system in any country. Then select country and propose set policies will move from its current state target state. Five indicators for evaluation are formulated. Then, Python crawler used capture data different countries under these five Internet. K-means++ clustering method classify into three classes. The significance specific results classification shown 4.2.1 Table3. according types countries,...

Abstract With the continuous development of science and technology, people for research artificial intelligence gradually in-depth, intelligent technology is widely used, can help to improve state life, people’s quality life. This paper will analyze computer identification explore its practical application in put forward some suggestions technology.

Block chain is a new application model integrating distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm and other computer technologies born under the background of rapid development Internet, its emergence has aroused attention various countries discussions in industries. The blockchain technology to accounting field been hot topic research exploration for scholars recent years. Its features decentralization, de-trust, immutability, traceability...