accuracy improvement
Recently Published Documents


TOTAL DOCUMENTS

1268
(FIVE YEARS 345)

H-INDEX

33
(FIVE YEARS 6)

2022 ◽  
pp. 1-10
Author(s):  
Daniel Trevino-Sanchez ◽  
Vicente Alarcon-Aquino

The need to detect and classify objects correctly is a constant challenge, being able to recognize them at different scales and scenarios, sometimes cropped or badly lit is not an easy task. Convolutional neural networks (CNN) have become a widely applied technique since they are completely trainable and suitable to extract features. However, the growing number of convolutional neural networks applications constantly pushes their accuracy improvement. Initially, those improvements involved the use of large datasets, augmentation techniques, and complex algorithms. These methods may have a high computational cost. Nevertheless, feature extraction is known to be the heart of the problem. As a result, other approaches combine different technologies to extract better features to improve the accuracy without the need of more powerful hardware resources. In this paper, we propose a hybrid pooling method that incorporates multiresolution analysis within the CNN layers to reduce the feature map size without losing details. To prevent relevant information from losing during the downsampling process an existing pooling method is combined with wavelet transform technique, keeping those details "alive" and enriching other stages of the CNN. Achieving better quality characteristics improves CNN accuracy. To validate this study, ten pooling methods, including the proposed model, are tested using four benchmark datasets. The results are compared with four of the evaluated methods, which are also considered as the state-of-the-art.


2022 ◽  
Author(s):  
Yukito Tsunoda ◽  
Toshihiko Mori ◽  
Tsuguchika Tabaru ◽  
Akira Oyama
Keyword(s):  

2022 ◽  
Author(s):  
Piao Su ◽  
Shu Liu ◽  
Hong Min ◽  
Yarui An ◽  
Chenglin Yan ◽  
...  

The rapid and accurate quantitative analysis of total iron (TFe) content in iron ores is extremely important in global iron ores trade. Due to the matrix effect among iron ores...


2021 ◽  
Vol 8 (4) ◽  
pp. 435-443
Author(s):  
Marcela Lascsáková

The focus of this paper aims at comparison of two prognostic numerical models with different strategies for accuracy improvement. To verify prediction performance of proposed models, the forecasts of aluminium stock exchanges on the London Metal Exchange were carried out as numerical solution of the Cauchy initial problem for the first-order ordinary differential equation. Two techniques for accuracy improvement were utilized, replacing the initial condition value by the nearest known stock exchange and a modification of the differential equation in solved Cauchy initial problem by means of two known initial values. We dealt with an idea of how different price development affected the accuracy of proposed strategies. With regard to obtained results, it was found that the prognoses obtained by using two known initial values were more increasing or decreasing than prognoses calculated by utilizing the initial condition drift. The strategy of a changing form of the differential equation in the Cauchy initial problem can be considered slightly more accurate. Faster increased prognoses were more advantageous especially at a steep price increase and within a price increase following the price decline. A moderate increase of the prognoses determined by the initial condition drift fit reasonably well a price fluctuation and a price decline following the price increase.


Author(s):  
Sai Kiran Cherupally ◽  
Jian Meng ◽  
Adnan Siraj Rakin ◽  
Shihui Yin ◽  
Injune Yeo ◽  
...  

Abstract We present a novel deep neural network (DNN) training scheme and RRAM in-memory computing (IMC) hardware evaluation towards achieving high robustness to the RRAM device/array variations and adversarial input attacks. We present improved IMC inference accuracy results evaluated on state-of-the-art DNNs including ResNet-18, AlexNet, and VGG with binary, 2-bit, and 4-bit activation/weight precision for the CIFAR-10 dataset. These DNNs are evaluated with measured noise data obtained from three different RRAM-based IMC prototype chips. Across these various DNNs and IMC chip measurements, we show that our proposed hardware noise-aware DNN training consistently improves DNN inference accuracy for actual IMC hardware, up to 8% accuracy improvement for the CIFAR-10 dataset. We also analyze the impact of our proposed noise injection scheme on the adversarial robustness of ResNet-18 DNNs with 1-bit, 2-bit, and 4-bit activation/weight precision. Our results show up to 6% improvement in the robustness to black-box adversarial input attacks.


Sign in / Sign up

Export Citation Format

Share Document