Theβ-Ga_(2)O_(3)films are prepared on polished Al_(2)O_(3)(0001)substrates by pulsed laser deposition at different oxygen partial *** influence of oxygen partial pressure on crystal structure,surface morphology,thick...
详细信息
Theβ-Ga_(2)O_(3)films are prepared on polished Al_(2)O_(3)(0001)substrates by pulsed laser deposition at different oxygen partial *** influence of oxygen partial pressure on crystal structure,surface morphology,thickness,optical properties,and photoluminescence properties are studied by x-ray diffraction(XRD),atomic force microscope(AFM),scanning electron microscope(SEM),spectrophotometer,and *** results of x-ray diffraction and atomic force microscope indicate that with the decrease of oxygen pressure,the full width at half maximum(FWHM)and grain size *** the increase of oxygen pressure,the thickness of the films first increases and then *** room-temperature UV-visible(UV-Vis)absorption spectra show that the bandgap of theβ-Ga_(2)O_(3)film increases from4.76 e V to 4.91 e V as oxygen pressure *** temperature photoluminescence spectra reveal that the emission band can be divided into four Gaussian bands centered at about 310 nm(~4.0 e V),360 nm(~3.44 e V),445 nm(~2.79 e V),and 467 nm(~2.66 e V),*** addition,the total photoluminescence intensity decreases with oxygen pressure increasing,and it is found that the two UV bands are related to self-trapped holes(STHs)at O1 sites and between two O2-s sites,respectively,and the two blue bands originate from V_(Ga)^(2-)at Ga1 tetrahedral *** photoluminescence mechanism of the films is also *** results will lay a foundation for investigating the Ga_(2)O_(3)film-based electronic devices.
Rockburst is a phenomenon where sudden,catastrophic failure of the rock mass occurs in underground deep regions or areas with high tectonic stress during the excavation *** disasters endanger the safety of people'...
详细信息
Rockburst is a phenomenon where sudden,catastrophic failure of the rock mass occurs in underground deep regions or areas with high tectonic stress during the excavation *** disasters endanger the safety of people's lives and property,national energy security,and social interests,so it is very important to accurately predict *** rockburst prediction has not been able to find an effective prediction method,and the study of the rockburst mechanism is facing a *** the development of artificial intelligence(AI)techniques in recent years,more and more experts and scholars have begun to introduce AI techniques into the study of the rockburst *** previous research,several scholars have attempted to summarize the application of AI techniques in rockburst ***,these studies either are not specifically focused on reviews of the application of AI techniques in rockburst prediction,or they do not provide a comprehensive *** on the advantages of extensive interdisciplinary research and a deep understanding of AI techniques,this paper conducts a comprehensive review of rockburst prediction methods leveraging AI ***,pertinent definitions of rockburst and its associated hazards are ***,the applications of both traditional prediction methods and those rooted in AI techniques for rockburst prediction are summarized,with emphasis placed on the respective advantages and disadvantages of each ***,the strengths and weaknesses of prediction methods leveraging AI are summarized,alongside forecasting future research trends to address existing challenges,while simultaneously proposing directions for improvement to advance the field and meet emerging demands effectively.
Due to a vast increase in security breaches and other violations, data hiding has become a vital measure for the protection of the integrity of an individual or a group. Image steganography is a perfect technique that...
详细信息
ISBN:
(数字)9798331516284
ISBN:
(纸本)9798331516291
Due to a vast increase in security breaches and other violations, data hiding has become a vital measure for the protection of the integrity of an individual or a group. Image steganography is a perfect technique that exists for the protection of data from getting altered and getting safely transmitted. Steganography serves as an excellent method to secure information as the unauthorized party barely recognizes the secret message or a text that is hidden behind the original or the cover image. In this project we have mainly focused on the Least Significant Bit (LSB) for hiding the texts behind the images and build our idea through a GUI. LSB is quite simple and an efficient algorithm in which the least significant bit that barely changes the quality or the build of the original image is changed and is replaced by the secret text that we want to communicate to the other users. Image preprocessing and generation of the output stego image is done by Pillow (PIL) and OpenCV in which image binary pixels are taken as input then they are altered with the secret text and then finally sent to the desired user.
Density based clustering (DBC) is an unsupervised learning method which is capable of identifying distinct groups of information. It works on the principle of instance-based variance maximization between distinct grou...
Density based clustering (DBC) is an unsupervised learning method which is capable of identifying distinct groups of information. It works on the principle of instance-based variance maximization between distinct groups, while minimizing variance between similar groups of data. DBC models are capable of detecting regions of similar concentrated instances, and discard sparse instances based on pre-set distance ratios. Thus, selection of this distance ratio must be done intelligently in order to perform effective density-based clustering. A wide range of models are proposed for this purpose, but most of them work in a context-dependent manner, thereby limiting their scalability. In order to remove this drawback, an augmented rule mining approach that utilizes a combination of Apriori and frequent pattern growth (FPGrowth) is proposed in this text. The proposed approach initially evaluates minimum support values for Apriori & FPGrowth models via variance-based analytical modelling. Rules for these models are combined using instance-based matching, and are used for parametric tuning of Density-Based Spatial Clustering of Applications with Noise (DBSCAN) model. Due to a combination of Apriori & FPGrowth a large number of rules are extracted in a dataset-independent manner, which assist in optimization of DBSCAN’s inter-class and intra-class variance metrics. The results are compared with various state-of-the-art clustering models, and cluster metrics are evaluated. It is observed that the proposed model outperforms DBSCAN, and other density-based clustering techniques in terms of cluster formation efficiency, scalability and computational complexity on multiple datasets.
The COVID-19 pandemic has greatly increased depression among adolescents. The current depression diagnosis process requires significant patient effort and can be costly. Prior research through passively collected data...
详细信息
ISBN:
(数字)9798350309652
ISBN:
(纸本)9798350309669
The COVID-19 pandemic has greatly increased depression among adolescents. The current depression diagnosis process requires significant patient effort and can be costly. Prior research through passively collected data has shown promising depression screening results but is limited by complex data collection and privacy concerns. In this research, we create multiple machine learning models to screen physiological data collected from Fitbit, a wearable biomarker, and depression screening surveys across 166 college students. The highest-scoring model on these physiological modalities achieved an F1-score of 0.92. Our research findings highlight the potential impact of digital technology development in current clinical practices.
With the significant development of large models in recent years, Large Vision-Language Models (LVLMs) have demonstrated remarkable capabilities across a wide range of multimodal understanding and reasoning tasks. Com...
详细信息
We study the limits and capability of public-data assisted differentially private (PA-DP) algorithms. Specifically, we focus on the problem of stochastic convex optimization (SCO) with either labeled or unlabeled publ...
ISBN:
(纸本)9798331314385
We study the limits and capability of public-data assisted differentially private (PA-DP) algorithms. Specifically, we focus on the problem of stochastic convex optimization (SCO) with either labeled or unlabeled public data. For complete/labeled public data, we show that any (ε, δ)-PA-DP has excess risk $\tilde{\Omega}\big(\min(\frac{1}{\sqrt{n_{\text{pub}}}},\frac{1}{\sqrt{n}}+\frac{\sqrt{d}}{n\epsilon} ) \big)$, where d is the dimension, npub is the number of public samples, npriv is the number of private samples, and n = npub + npriv. These lower bounds are established via our new lower bounds for PA-DP mean estimation, which are of a similar form. Up to constant factors, these lower bounds show that the simple strategy of either treating all data as private or discarding the private data, is optimal. We also study PA-DP supervised learning with unlabeled public samples. In contrast to our previous result, we here show novel methods for leveraging public data in private supervised learning. For generalized linear models (GLM) with unlabeled public data, we show an efficient algorithm which, given Õ(nprivε) unlabeled public samples, achieves the dimension independent rate $\tilde{O}\big(\frac{1}{\sqrt{{n_{\text{priv}}}}} + \frac{1}{\sqrt{{n_{\text{priv}}}\epsilon}}\big)$. We develop new lower bounds for this setting which shows that this rate cannot be improved with more public samples, and any fewer public samples leads to a worse rate. Finally, we provide extensions of this result to general hypothesis classes with finite fat-shattering dimension with applications to neural networks and non-Euclidean geometries.
Recommender systems play a critical role in personalizing content, but integrating multiple data modalities and addressing dynamic user preferences presents significant challenges. This paper introduces the Adaptive M...
详细信息
ISBN:
(数字)9798331537579
ISBN:
(纸本)9798331537586
Recommender systems play a critical role in personalizing content, but integrating multiple data modalities and addressing dynamic user preferences presents significant challenges. This paper introduces the Adaptive Multi-Modal Generative Recommender (AMGR) framework, which combines text, image, and video data to enhance recommendation accuracy. By leveraging advanced generative models such as GANs, VAEs, and transformers, AMGR not only suggests content based on user history but also generates personalized content that aligns with evolving preferences. To address the complexities of real-time adaptation, AMGR utilizes reinforcement learning strategies, balancing exploration and exploitation of content. The framework also tackles key technical and ethical challenges, including computational efficiency, multi-modal data fusion, fairness, and privacy. Potential solutions such as federated learning and bias mitigation techniques are explored to ensure responsible and scalable deployment. This paper outlines the proposed methodology, its applications, and the necessary safeguards for AMGR's real-world implementation, contributing to the development of adaptive, fair, and transparent recommendation systems.
Broad Learning System (BLS), characterized by its flat architecture as an efficient neural network, has garnered significant interest for its benefits in terms of training velocity and network scalability. However, BL...
详细信息
ISBN:
(数字)9798331529482
ISBN:
(纸本)9798331529499
Broad Learning System (BLS), characterized by its flat architecture as an efficient neural network, has garnered significant interest for its benefits in terms of training velocity and network scalability. However, BLS does not perform well when dealing with imbalanced classification problems. First, conventional BLS neglects the intrinsic logical information contained in the original input data, which may be critical for classification and prediction. Second, BLS fails to emphasize the significance of minority classes in imbalanced data scenarios. In this paper, we propose a feature fusion-based weighted broad learning system. Unlike traditional BLS, we combine the original features and broad features as fused features for training. Additionally, we introduce a weight generation mechanism that assigns higher weights to minority classes, enhancing the model’s focus on these classes. A closed solution is available for the issue, and the transformation matrix can be effectively resolved. The experimental results on multiple imbalanced datasets show that our method is superior to other imbalanced methods.
Deep learning (DL) has made significant advancements in tomographic imaging, particularly in low-dose computed tomography (LDCT) denoising. A recent trend involves servers training powerful models with enormous self-c...
详细信息
暂无评论