Brain tumors are ranked highly among the leading causes of cancer-related fatalities. Precise segmentation and quantitative assessment of brain tumors are crucial for effective diagnosis and treatment planning. Howeve...
详细信息
The global navigation satellite system-based technology has inherent limitations due to its reliance on radio signals. In contrast, visual localization operates independently of radio communication, presenting a viabl...
详细信息
In recent years, deep neural networks have achieved remarkable accuracy in computer vision tasks. With inference time being a crucial factor, particularly in dense prediction tasks such as semantic segmentation, knowl...
详细信息
The rapid growth of digital data necessitates advanced natural language processing(NLP)models like BERT(Bidi-rectional Encoder Representations from Transformers),known for its superior performance in text ***,BERT’s ...
详细信息
The rapid growth of digital data necessitates advanced natural language processing(NLP)models like BERT(Bidi-rectional Encoder Representations from Transformers),known for its superior performance in text ***,BERT’s size and computational demands limit its practicality,especially in resource-constrained *** research compresses the BERT base model for Bengali emotion classification through knowledge distillation(KD),pruning,and quantization *** Bengali being the sixth most spoken language globally,NLP research in this area is *** approach addresses this gap by creating an efficient BERT-based model for Bengali *** have explored 20 combinations for KD,quantization,and pruning,resulting in improved speedup,fewer parameters,and reduced memory *** best results demonstrate significant improvements in both speed and *** instance,in the case of mBERT,we achieved a 3.87×speedup and 4×compression ratio with a combination of Distil+Prune+Quant that reduced parameters from 178 to 46 M,while the memory size decreased from 711 to 178 *** results offer scalable solutions for NLP tasks in various languages and advance the field of model compression,making these models suitable for real-world applications in resource-limited environments.
The disappearance of Indigenous languages results in a decrease in cultural diversity, hence making the preservation of these languages extremely important. Conventional methods of documentation are lengthy, and the p...
详细信息
Small parasitic Hemipteran insects known as bedbugs(Cimicidae)feed on warm-blooded mammal’s *** most famous member of this family is the Cimex lectularius or common *** current paper proposes a novel swarm intelligen...
详细信息
Small parasitic Hemipteran insects known as bedbugs(Cimicidae)feed on warm-blooded mammal’s *** most famous member of this family is the Cimex lectularius or common *** current paper proposes a novel swarm intelligence optimization algorithm called the Bedbug Meta-Heuristic Algorithm(BMHA).The primary inspiration for the bedbug algorithm comes from the static and dynamic swarming behaviors of bedbugs in *** two main stages of optimization algorithms,exploration,and exploitation,are designed by modeling bedbug social interaction to search for *** proposed algorithm is benchmarked qualitatively and quantitatively using many test functions including *** results of evaluating BMHA prove that this algorithm can improve the initial random population for a given optimization problem to converge towards global optimization and provide highly competitive results compared to other well-known optimization *** results also prove the new algorithm's performance in solving real optimization problems in unknown search *** achieve this,the proposed algorithm has been used to select the features of fake news in a semi-supervised manner,the results of which show the good performance of the proposed algorithm in solving problems.
Indoor positioning is a key technology in today’s intelligent environments,and it plays a crucial role in many application *** paper proposed an unscented Kalman filter(UKF)based on the maximum correntropy criterion(...
详细信息
Indoor positioning is a key technology in today’s intelligent environments,and it plays a crucial role in many application *** paper proposed an unscented Kalman filter(UKF)based on the maximum correntropy criterion(MCC)instead of the minimummean square error criterion(MMSE).This innovative approach is applied to the loose coupling of the Inertial Navigation System(INS)and Ultra-Wideband(UWB).By introducing the maximum correntropy criterion,the MCCUKF algorithm dynamically adjusts the covariance matrices of the system noise and the measurement noise,thus enhancing its adaptability to diverse environmental localization *** in the presence of non-Gaussian noise,especially heavy-tailed noise,the MCCUKF exhibits superior accuracy and robustness compared to the traditional *** method initially generates an estimate of the predicted state and covariance matrix through the unscented transform(UT)and then recharacterizes the measurement information using a nonlinear regression method at the cost of ***,the state and covariance matrices of the filter are updated by employing the unscented transformation on the measurement ***,to mitigate the influence of non-line-of-sight(NLOS)errors positioning accuracy,this paper proposes a k-medoid clustering algorithm based on bisection k-means(Bikmeans).This algorithm preprocesses the UWB distance measurements to yield a more precise position *** results demonstrate that MCCUKF is robust to the uncertainty of UWB and realizes stable integration of INS and UWB systems.
The use of all samples in the optimization process does not produce robust results in datasets with label *** the gradients calculated according to the losses of the noisy samples cause the optimization process to go ...
详细信息
The use of all samples in the optimization process does not produce robust results in datasets with label *** the gradients calculated according to the losses of the noisy samples cause the optimization process to go in the wrong *** this paper,we recommend using samples with loss less than a threshold determined during the optimization,instead of using all samples in the *** proposed method,Adaptive-k,aims to exclude label noise samples from the optimization process and make the process *** noisy datasets,we found that using a threshold-based approach,such as Adaptive-k,produces better results than using all samples or a fixed number of low-loss samples in the *** the basis of our theoretical analysis and experimental results,we show that the Adaptive-k method is closest to the performance of the Oracle,in which noisy samples are entirely removed from the ***-k is a simple but effective *** does not require prior knowledge of the noise ratio of the dataset,does not require additional model training,and does not increase training time *** the experiments,we also show that Adaptive-k is compatible with different optimizers such as SGD,SGDM,and *** code for Adaptive-k is available at GitHub.
This study proposes a real-time integrated framework for LiDAR-based object tracking in autonomous driving environments. Advancements in LiDAR sensors are increasing point cloud data collection, leading to a demand fo...
详细信息
Accurate skin disease detection is one of the most challenging tasks due to high-class imbalance and limited labeled datasets. Recently Deep Convolutional Neural Network (DCNN) with ensemble learning has achieved sign...
详细信息
Accurate skin disease detection is one of the most challenging tasks due to high-class imbalance and limited labeled datasets. Recently Deep Convolutional Neural Network (DCNN) with ensemble learning has achieved significant popularity in skin cancer classification. However, implementing DCNN models with ensemble learning is not feasible for deployment on portable diagnostic devices due to the limitation in computing resources and computing time. This paper proposes a Channel Attention and Adaptive Class Balanced Focal Loss function based lightweight Deep CNN model (CACBL-Net) for handling the issues of data imbalance and limited computing resources of portable diagnostic devices, such as mobile phones or tablets. Channel attention explores interdependencies between channels by recalibrating channel-wise feature responses. To deal with the issue of high-class imbalance, the proposed method used an adaptive class balance focal loss function which can quickly concentrate the model on complex cases while automatically downweighting the contribution of easy examples during training. The proposed CACBL-Net is validated on three popular skin cancer datasets which are HAM-10000, PAD-UFES-20, and MED-NODE. Dermoscopic, non-dermoscopic and smartphone images are taken from all three datasets for experimental work. The quantitative findings indicate that the proposed CACBL-Net model achieved a sensitivity of 90.60%, 91.88%, and 91.31% for the HAM-10000, PAD-UFES-20, and MED-NODE datasets, respectively. Additionally, the average prediction time per patient was recorded at 0.006, 0.010, and 0.011 s. These results demonstrate superior performance compared to other state-of-the-art deep learning models. The experimental finding suggested that the proposed method can achieve a significant performance at a low cost of computational resources and inference time, which makes it potentially feasible for deployment in portable diagnostic devices for automated diagnosis of skin lesions.
暂无评论