Recently, deep convolutional neural networks (DCNNs) have achieved remarkable results in image classification tasks. Despite convolutional networks’ great successes, their training process relies on a large amount of...Recently, deep convolutional neural networks (DCNNs) have achieved remarkable results in image classification tasks. Despite convolutional networks’ great successes, their training process relies on a large amount of data prepared in advance, which is often challenging in real-world applications, such as streaming data and concept drift. For this reason, incremental learning (continual learning) has attracted increasing attention from scholars. However, incremental learning is associated with the challenge of catastrophic forgetting: the performance on previous tasks drastically degrades after learning a new task. In this paper, we propose a new strategy to alleviate catastrophic forgetting when neural networks are trained in continual domains. Specifically, two components are applied: data translation based on transfer learning and knowledge distillation. The former translates a portion of new data to reconstruct the partial data distribution of the old domain. The latter uses an old model as a teacher to guide a new model. The experimental results on three datasets have shown that our work can effectively alleviate catastrophic forgetting by a combination of the two methods aforementioned.展开更多
Blasting is one of the most important operations in the mining projects that has effective role in the whole operation physically and economically. Unsuitable blasting pattern may lead to unwanted events such as poor ...Blasting is one of the most important operations in the mining projects that has effective role in the whole operation physically and economically. Unsuitable blasting pattern may lead to unwanted events such as poor fragmentation, back break and fly rock. Multi attribute decision making(MADM) can be useful method for selecting the most appropriate blasting pattern among previously performed patterns. In this work, initially, from various already performed patterns, efficient and inefficient patterns are determined using data envelopment analysis(DEA). In the second step, after weighting impressive attributes using experts' opinion, elimination Et choice translating reality(ELECTRE) was used for ranking the efficient patterns and recognizing the most appropriate pattern in the Sungun Copper Mine, Iran. According to the obtained results, blasting pattern with the hole diameter of 15.24 cm, burden of 3 m, spacing of 4 m and stemming of 3.2 m has selected as the best pattern and has selected for future operation.展开更多
文摘Recently, deep convolutional neural networks (DCNNs) have achieved remarkable results in image classification tasks. Despite convolutional networks’ great successes, their training process relies on a large amount of data prepared in advance, which is often challenging in real-world applications, such as streaming data and concept drift. For this reason, incremental learning (continual learning) has attracted increasing attention from scholars. However, incremental learning is associated with the challenge of catastrophic forgetting: the performance on previous tasks drastically degrades after learning a new task. In this paper, we propose a new strategy to alleviate catastrophic forgetting when neural networks are trained in continual domains. Specifically, two components are applied: data translation based on transfer learning and knowledge distillation. The former translates a portion of new data to reconstruct the partial data distribution of the old domain. The latter uses an old model as a teacher to guide a new model. The experimental results on three datasets have shown that our work can effectively alleviate catastrophic forgetting by a combination of the two methods aforementioned.
文摘Blasting is one of the most important operations in the mining projects that has effective role in the whole operation physically and economically. Unsuitable blasting pattern may lead to unwanted events such as poor fragmentation, back break and fly rock. Multi attribute decision making(MADM) can be useful method for selecting the most appropriate blasting pattern among previously performed patterns. In this work, initially, from various already performed patterns, efficient and inefficient patterns are determined using data envelopment analysis(DEA). In the second step, after weighting impressive attributes using experts' opinion, elimination Et choice translating reality(ELECTRE) was used for ranking the efficient patterns and recognizing the most appropriate pattern in the Sungun Copper Mine, Iran. According to the obtained results, blasting pattern with the hole diameter of 15.24 cm, burden of 3 m, spacing of 4 m and stemming of 3.2 m has selected as the best pattern and has selected for future operation.
