Open-world Semi-Supervised Learning (OSSL) is a realistic and challenging task, aiming to classify unlabeled samples from both seen and novel classes using partially labeled samples from the seen classes. Previous wor...
Open-world Semi-Supervised Learning (OSSL) is a realistic and challenging task, aiming to classify unlabeled samples from both seen and novel classes using partially labeled samples from the seen classes. Previous works typically explore the relationship of samples as priors on the pre-defined single-granularity labels to help novel class recognition. In fact, classes follow a taxonomy and samples can be classified at multiple levels of granularity, which contains more underlying relationships for supervision. We thus argue that learning with single-granularity labels results in sub-optimal representation learning and inaccurate pseudo labels, especially with unknown classes. In this paper, we take the initiative to explore and propose a uniformed framework, called Taxonomic context prIors Discovering and Aligning (TIDA), which exploits the relationship of samples under various granularity. It allows us to discover multi-granularity semantic concepts as taxonomic context priors (i.e., sub-class, target-class, and super-class), and then collaboratively leverage them to enhance representation learning and improve the quality of pseudo labels. Specifically, TIDA comprises two components: i) A taxonomic context discovery module that constructs a set of hierarchical prototypes in the latent space to discover the underlying taxonomic context priors; ii) A taxonomic context-based prediction alignment module that enforces consistency across hierarchical predictions to build the reliable relationship between classes among various granularity and provide additions supervision. We demonstrate that these two components are mutually beneficial for an effective OSSL framework, which is theoretically explained from the perspective of the EM algorithm. Extensive experiments on seven commonly used datasets show that TIDA can significantly improve the performance and achieve a new state of the art. The source codes are publicly available at https://***/rain305f/TIDA.
Autonomous take-off and landing capabilities are crucial in UAV vision-based missions, ensuring adaptive navigation, especially in challenging environments where realtime identification and interaction with a variety ...
详细信息
ISBN:
(数字)9798331513283
ISBN:
(纸本)9798331513290
Autonomous take-off and landing capabilities are crucial in UAV vision-based missions, ensuring adaptive navigation, especially in challenging environments where realtime identification and interaction with a variety of landing platforms are required. In this context, this paper presents a servo-visual controller that uses pattern detection and color segmentation techniques to identify take-off/landing platforms and estimate their current orientation. The proposed system was subjected to experimental validation with four platforms positioned in different orientations, heights, and positions, demonstrating its versatility in various conditions. Our study addresses the Flying Robots Trial League challenge, which emulates mapping and inspection tasks in offshore platforms.
Space has increasingly attracted the attention of governments, large industries, and universities. One of the most popular strategies in recent years has been the adoption of nanosatellites to fulfill different missio...
Space has increasingly attracted the attention of governments, large industries, and universities. One of the most popular strategies in recent years has been the adoption of nanosatellites to fulfill different missions, which can work alone or in constellations. Universities stand out among the agents launching nanosatellites, with more than 600 launches until 2022. Given the growth of entities that control space missions, it is necessary to implement new methods for communication between control and satellite to accelerate data transmission and provide a high-security degree. Our work proposes a consortium archi-tecture between Ground Stations (GSs) so that a GS as a Service (GSaaS) works with low cost, reliability, and resource sharing. We simulated a nanosatellite mission in Low Earth Orbit (LEO) with MATLAB to obtain the parameters of average communication time, propagation loss, and at which angles the communication would be most affected by atmospheric phenomena. Then, we implement business rules for communication between GS and satellites using smart contract concepts. We set up a blockchain to provide the decentralization infrastructure and created a web service to provide a communication API between nanosatellite and blockchain. We simulated the firmware update process, showing that the nanosatellite took around 20 minutes to request all 32-byte fragments of 301 Kb firmware. Considering the time interval that the communication window between GS and nanosatellite remains active, the entire firmware transmission takes two to three communication slots. However, the transmission time is drastically reduced in a scenario with two or more GSs. Furthermore, the GSaaS decentralized infrastructure allows the consortium of GSs to communicate agnostically with the satellites, preserving firmware privacy due to the cryptography used in blockchain transactions
This paper presents an optical coherence tomography (OCT) system in conjunction with a novel image reconstruction technique employed for in vitro imaging of human teeth. The primary goal is to enhance the signal-to-no...
This paper presents an optical coherence tomography (OCT) system in conjunction with a novel image reconstruction technique employed for in vitro imaging of human teeth. The primary goal is to enhance the signal-to-noise ratio (SNR) in the obtained images. The study entails a comparative analysis between the conventional Fast Fourier Transform (FFT) OCT image reconstruction method and a newly introduced scaled nonuniform discrete Fourier transform (NDFT) approach. The findings reveal that the NDFT method consistently delivers superior results in terms of peak signal-to-noise ratio (PSNR) and overall image quality, even when dealing with redundant and nonuniform frequency domain samples. In light of these results, this paper concludes that integrating NDFT into OCT procedures has the potential to significantly enhance the quality of image reconstructions, thereby fostering its broader application in the field of dental imaging.
This paper describes the low-cost manufacturing process of an evanescent wave fiber sensor platform that allows the etching of the fiber in hydrofluoric acid with the proposed 3D-printed fiber holder in an acid resist...
详细信息
ISBN:
(数字)9786589532026
ISBN:
(纸本)9798350362725
This paper describes the low-cost manufacturing process of an evanescent wave fiber sensor platform that allows the etching of the fiber in hydrofluoric acid with the proposed 3D-printed fiber holder in an acid resistant material. The fiber holder is practical for allowing several processing procedures during its use and sensor fabrication, allowing sensor optical and electrical operation. In this work, some aspects of its characterization are presented in the optical domain showing the excitation of surface plasmons in the visible range, and also indication of its electrical operation of the thin-film as electrode.
This paper proposes a reconfiguration for a single-phase, double conversion uninterruptible power supply (UPS). With the addition of two static switches to the original circuit, the UPS mitigates the ripple in the cur...
This paper proposes a reconfiguration for a single-phase, double conversion uninterruptible power supply (UPS). With the addition of two static switches to the original circuit, the UPS mitigates the ripple in the current, through battery discharge performed by two legs. Other than these switches, the circuit topology is preserved. The preliminary results show a satisfactory level of ripples for the inductor current and suitable power quality for the output voltage.
This thesis presents methods and approaches to image color correction, color enhancement, and color editing. To begin, we study the color correction problem from the standpoint of the camera's image signal process...
详细信息
Crop Yield Analysis and Prediction is a fast-expanding discipline that is critical for optimizing agricultural methods. A lack of trustworthy data is one of the challenges in estimating crop yields. We develop predict...
Crop Yield Analysis and Prediction is a fast-expanding discipline that is critical for optimizing agricultural methods. A lack of trustworthy data is one of the challenges in estimating crop yields. We develop predictive models for 22 different fruits and vegetables data. The goals of this study are to create accurate and interpretable crop recommendation models. We used multiple machine learning (ML) models for multi-class crop production prediction to fulfill our research goal. We thoroughly examined the influence of climate and nutrient factors on crop yield, considering their complex interactions. To improve the dataset, augmented data techniques were applied. Configuring the parameters and fine-tuning the hyperparameters is our technique to increase the model performance. Furthermore, we employ explainable artificial intelligence (XAI) techniques and interpretability tools like Shapley Additive exPlanations (SHAP) to improve the interpretability of our prediction model. Our findings reveal that the XGBoost model has the best performance model with 99.86% accuracy, followed by SVM Poly Kernel with 99.32% and Random Forest with 98.82%. Feature selection and analysis are emphasized, particularly in regional agricultural contexts. This study contributes to the creation of accurate and interpretable crop recommendation models while also addressing the issue of untrustworthy data, providing useful insights for optimizing agricultural practices.
Navigating the complex legal and regulatory landscape requires a sophisticated platform that is not only comprehensive but also user-friendly and enables seamless analysis and document comparison in the legal realm. T...
Navigating the complex legal and regulatory landscape requires a sophisticated platform that is not only comprehensive but also user-friendly and enables seamless analysis and document comparison in the legal realm. To address this challenging requirement, this research is dedicated to the user experience (UX) design of state-of the-art cloud-native mobile applications specifically designed for legal and regulatory harmonization. The methodology used is based on design thinking principles and takes advantage of the benefits of cloud computing. These include, but are not limited to, scalability, reliability, and security. The result is a solution that is not only powerful but also cost-effective. The core of this application follows a user-centered design philosophy and involves potential users closely in the design process. A comprehensive evaluation is performed using the System Usability Scale (SUS) methodology to measure the effectiveness and usability of the application. The results of this comprehensive approach are convincing. The application has an impressive 95.5% usage rate, indicating strong user engagement. Furthermore, user satisfaction ratings are particularly high and transformative impact of this application in a complex environment of legal harmonization. Through its innovative design, seamless functionality, and user-centric approach, this application has proven to be not only a technical solution, but also a catalyst for positive change in the legal and regulatory harmonization process.
Batik is an Indonesian world cultural heritage. Batik consists of many kinds of patterns depending on where the batik comes from, Batik-making techniques continue to develop along with technology development. Among th...
Batik is an Indonesian world cultural heritage. Batik consists of many kinds of patterns depending on where the batik comes from, Batik-making techniques continue to develop along with technology development. Among the batik making techniques that are widely used are hand-written, stamping, and printing. Batik motifs have been widely used as research material, especially in the field of artificial intelligence. The diverse appearance of batik motifs has attracted many researchers to carry out research on making synthetic batik patterns, one of which uses a Generative Adversarial Network. This paper presents a synthetic batik pattern model based on the Wasserstein Generative Adversarial Network with Gradient Penalty. This model has been proven to create new synthetic batik patterns quite well and almost identical with images provided in the dataset, with the notes if the dataset provided is large.
暂无评论