Filter Training Research Articles

A Fusion Method for Multisource Land Cover Products Based on Superpixels and Statistical Extraction for Enhancing Resolution and Improving Accuracy

The discrepancies in existing land cover data are relatively high, indicating low local precision and application limitations. Multisource data fusion is an effective way to solve this problem; however, the fusion procedure often requires resampling to unify the spatial resolution, causing a lower spatial resolution. To solve this problem, this study proposes a multisource product fusion mapping method of filtering training samples and product correction at a fine resolution. Based on the Superpixel algorithm, principal component analysis (PCA), and statistical extraction techniques, combined with the Google Earth Engine (GEE) platform, reliable land cover data were acquired. GEE and machine-learning algorithms correct the unreliable information of multiple products into a new land cover fusion result. Compared to the common method of extracting consistent pixels from existing products, our proposed method effectively removes nearly 38.75% of them, with a high probability of classification error. The overall accuracy of fusion in this study reached 85.80%, and the kappa coefficient reached 0.82, with an overall accuracy improvement of 11.75–24.17% and a kappa coefficient improvement of 0.16 to 0.3 compared to other products. For existing single-category products, we corrected the phenomenon of overinterpretation in inconsistent areas; the overall accuracy improvement ranged from 2.99% to 20.71%, while the kappa coefficient improvement ranged from 0.22 to 0.56. Thus, our proposed method can combine information from multiple products and serve as an effective method for large areas and even as a global land cover fusion product.

Decision-Making Deficits in Elderly Can Be Alleviated by Attention Training.

Decision-making is an important everyday function that deteriorates during normal aging. Here, we asked whether value-based decision-making can be improved in the elderly by cognitive training. We compared the effects of two training regimens on the performance in the Iowa Gambling Task (IGT), a real-life decision-making simulation task. Elderly participants (age 62–75 years) were randomized into three matched groups. The filter training (FT) group performed a selective attention task and the memory training (MT) group performed a memory storage task on five consecutive days. The control group (CG) did not perform another task besides the IGT. Only the FT group showed an improvement in IGT performance over the five days—the overall gain rose and the prominent deck B phenomenon decreased. The latter refers to the selection of cards associated with high gains and rare losses, which are nevertheless a disadvantageous choice as the frequent losses lead to a negative net outcome. As the deck B phenomenon has been associated with impaired cognitive abilities in aging, the positive effect of FT here is of special importance. In sum, attention training seems superior in improving decision-making in the elderly.

An Approach to Overcome Occlusions in Visual Tracking: By Occlusion Estimating Agency and Self-Adapting Learning Rate for Filter's Training

Visual tracking methods have been successful in recent years. Correlation filter (CF) based methods significantly advanced state-of-the-art tracking. The advancement in CF tracking performance is predominantly attributed to powerful features and sophisticated online learning formulations. However, there would be trouble if the tracker indiscriminately learned samples. Particularly, when the target is severely occluded or out-of-view, the tracker will continuously learn the wrong information, resulting target loss in the following frames. In this study, aiming to avoid incorrect training when occlusions occur, we propose a regional color histogram-based occlusion estimating agency (RCHBOEA), which estimates the occlusion level and then instructs, based on the result, the tracker to work in one of two modes: normal or lost. In the normal mode, an occlusion level-based self-adopting learning rate is used for tracker training. In the lost mode, the tracker pauses its training and conducts a search and recapture strategy on a wider searching area. Our method can easily complement CF-based trackers. In our experiments, we employed four CF-based trackers as a baseline: discriminative CFs (DCF), kernelized CFs (KCF), background-aware CFs (BACF), and efficient convolution operators for tracking: hand-crafted feature version (ECO_HC). We performed extensive experiments on the standard benchmarks: VIVID, OTB50, and OTB100. The results demonstrated that combined with RCHBOEA, the trackers achieved a remarkable improvement.

DCF: A Dataflow-Based Collaborative Filtering Training Algorithm

Emerging recommender systems often adopt collaborative filtering techniques to improve the recommending accuracy. Existing collaborative filtering techniques are implemented with either alternating least square algorithm or gradient descent (GD) algorithm. However, both of the two algorithms are not scalable because ALS suffers from high computation complexity and GD suffers from severe synchronization problem and tremendous data movement. To solve the above problems, we proposed a Dataflow-based Collaborative Filtering (DCF) algorithm. More specifically, DCF exploits fine-grain asynchronous feature of dataflow model to minimize synchronization overhead; leverages mini-batch technique to reduce computation and communication complexities; uses dummy edge and multicasting techniques to avoid fine-grain overhead of dependency checking and reduce data movement. By utilizing all the above techniques, DCF is able to significantly improve the performance of collaborative filtering. Our experiment on a cluster with one master node and ten slave nodes show that DCF achieves 23 $$\times $$ speedup over ALS on Spark and 18 $$\times $$ speedup over GD on Graphlab in public datasets.

Attentional Filter Training but Not Memory Training Improves Decision-Making.

Decision-making has a high practical relevance for daily performance. Its relation to other cognitive abilities such as executive control and memory is not fully understood. Here we asked whether training of either attentional filtering or memory storage would influence decision-making as indexed by repetitive assessments of the Iowa Gambling Task (IGT). The IGT was developed to assess and simulate real-life decision-making (Bechara et al., 2005). In this task, participants gain or lose money by developing advantageous or disadvantageous decision strategies. On five consecutive days we trained 29 healthy young adults (20–30 years) either in working memory (WM) storage or attentional filtering and measured their IGT scores after each training session. During memory training (MT) subjects performed a computerized delayed match-to-sample task where two displays of bars were presented in succession. During filter training (FT) participants had to indicate whether two simultaneously presented displays of bars matched or not. Whereas in MT the relevant target stimuli stood alone, in FT the targets were embedded within irrelevant distractors (bars in a different color). All subjects within each group improved their performance in the trained cognitive task. For the IGT, we observed an increase over time in the amount of money gained in the FT group only. Decision-making seems to be influenced more by training to filter out irrelevant distractors than by training to store items in WM. Selective attention could be responsible for the previously noted relationship between IGT performance and WM and is therefore more important for enhancing efficiency in decision-making.

Training of Attentional Filtering, but Not of Memory Storage, Enhances Working Memory Efficiency by Strengthening the Neuronal Gatekeeper Network.

Memory training (MT) in older adults with memory deficits often leads to frustration and, therefore, is usually not recommended. Here, we pursued an alternative approach and looked for transfer effects of 1-week attentional filter training (FT) on working memory performance and its neuronal correlates in young healthy humans. The FT effects were compared with pure MT, which lacked the necessity to filter out irrelevant information. Before and after training, all participants performed an fMRI experiment that included a combined task in which stimuli had to be both filtered based on color and stored in memory. We found that training induced processing changes by biasing either filtering or storage. FT induced larger transfer effects on the untrained cognitive function than MT. FT increased neuronal activity in frontal parts of the neuronal gatekeeper network, which is proposed to hinder irrelevant information from being unnecessarily stored in memory. MT decreased neuronal activity in the BG part of the gatekeeper network but enhanced activity in the parietal storage node. We take these findings as evidence that FT renders working memory more efficient by strengthening the BG-prefrontal gatekeeper network. MT, on the other hand, simply stimulates storage of any kind of information. These findings illustrate a tight connection between working memory and attention, and they may open up new avenues for ameliorating memory deficits in patients with cognitive impairments.