Search for: All records

Accurate and timely large-scale paddy rice maps with remote sensing are essential for crop monitoring and management and are used for assessing its impacts on food security, water resource management, and transmission of zoonotic infectious diseases. Optical image-based paddy rice mapping studies employed the unique spectral feature during the flooding/transplanting period of paddy rice. However, the lack of high-quality observations during the flooding/transplanting stage caused by rain and clouds and spectral similarity between paddy rice and natural wetlands often introduce errors in paddy rice identification, especially in paddy rice and wetland coexistent areas. In this study, we used a knowledge-based algorithm and time series observation from optical images (Sentinel-2 and Landsat 7/8) and microwave images (Sentinel-1) to address these issues. The final 10-m paddy rice map had user’s accuracy, producer’s accuracy, F1-score, and overall accuracy of 0.91 ± 0.004, 0.74 ± 0.010, 0.82, and 0.98 ± 0.001 (± value is the standard error), respectively. Over half (62.0%) of the paddy rice pixels had a confidence level of 1 (detected by both optical images and microwave images), while 38.0% had a confidence level of 0.5 (detected by either optical images or microwave images). The estimated paddy rice area in northeast China for 2020 was 60.83 ± 0.86 × 103 km2. Provincial and municipal rice areas in our data set agreed well with other existing paddy rice data sets and the Agricultural Statistical Yearbooks. These findings indicate that knowledge-based paddy rice mapping algorithms and a combination of optical and microwave images hold great potential for timely and frequently accurate paddy rice mapping in large-scale complex landscapes. 

Highly pathogenic avian influenza viruses (HPAIV) persistently threaten wild waterfowl, domestic poultry, and public health. The East Asian–Australasian Flyway plays a crucial role in HPAIV dynamics due to its large populations of migratory waterfowl and poultry. Over recent decades, this flyway has undergone substantial landscape changes, including both losses and gains of waterfowl habitats. These changes can affect waterfowl distributions, increase contact with poultry, and consequently alter ecological conditions that favor avian influenza virus (AIV) evolution. However, limited research has assessed these likely impacts. Here, we integrated empirical data and an individual-based model to simulate AIV transmission in migratory waterfowl and domestic poultry, including wild-to-poultry spillover and reassortment dynamics in poultry, across landscapes representing the years 2000 and 2015. We used the reassortment incidence as a proxy for ecological and transmission conditions that support viral diversification and the emergence of novel subtypes. Our simulations show that landscape change reshaped the waterfowl distribution, facilitated bird aggregation at improved habitats, increased coinfection, and raised reassortment rate by 1,593%, indicating a substantially higher potential for viral diversification and emergence. Model-generated risk maps show expanded and increased reassortment risk in southeastern China, the Yellow River Basin, and northeastern China. These findings suggest the importance of landscape change as a driver of potential AIV diversification and subtype emergence. This underscores the need for interdisciplinary approaches that integrate landscape dynamics, host movement, and viral evolution to better assess and mitigate future risk. 

Neuromorphic computing systems promise high energy efficiency and low latency. In particular, when integrated with neuromorphic sensors, they can be used to produce intelligent systems for a broad range of applications. An event‐based camera is such a neuromorphic sensor, inspired by the sparse and asynchronous spike representation of the biological visual system. However, processing the event data requires either using expensive feature descriptors to transform spikes into frames, or using spiking neural networks (SNNs) that are expensive to train. In this work, a neural network architecture is proposed, reservoir nodes‐enabled neuromorphic vision sensing network (RN‐Net), based on dynamic temporal encoding by on‐sensor reservoirs and simple deep neural network (DNN) blocks. The reservoir nodes enable efficient temporal processing of asynchronous events by leveraging the native dynamics of the node devices, while the DNN blocks enable spatial feature processing. Combining these blocks in a hierarchical structure, the RN‐Net offers efficient processing for both local and global spatiotemporal features. RN‐Net executes dynamic vision tasks created by event‐based cameras at the highest accuracy reported to date at one order of magnitude smaller network size. The use of simple DNN and standard backpropagation‐based training rules further reduces implementation and training costs. 

Analog compute‐in‐memory (CIM) systems are promising candidates for deep neural network (DNN) inference acceleration. However, as the use of DNNs expands, protecting user input privacy has become increasingly important. Herein, a potential security vulnerability is identified wherein an adversary can reconstruct the user's private input data from a power side‐channel attack even without knowledge of the stored DNN model. An attack approach using a generative adversarial network is developed to achieve high‐quality data reconstruction from power leakage measurements. The analyses show that the attack methodology is effective in reconstructing user input data from power leakage of the analog CIM accelerator, even at large noise levels and after countermeasures. To demonstrate the efficacy of the proposed approach, an example of CIM inference of U‐Net for brain tumor detection is attacked, and the original magnetic resonance imaging medical images can be successfully reconstructed even at a noise level of 20% standard deviation of the maximum power signal value. This study highlights a potential security vulnerability in emerging analog CIM accelerators and raises awareness of needed safety features to protect user privacy in such systems. 

Abstract The constant drive to achieve higher performance in deep neural networks (DNNs) has led to the proliferation of very large models. Model training, however, requires intensive computation time and energy. Memristor‐based compute‐in‐memory (CIM) modules can perform vector‐matrix multiplication (VMM) in place and in parallel, and have shown great promises in DNN inference applications. However, CIM‐based model training faces challenges due to non‐linear weight updates, device variations, and low‐precision. In this work, a mixed‐precision training scheme is experimentally implemented to mitigate these effects using a bulk‐switching memristor‐based CIM module. Low‐precision CIM modules are used to accelerate the expensive VMM operations, with high‐precision weight updates accumulated in digital units. Memristor devices are only changed when the accumulated weight update value exceeds a pre‐defined threshold. The proposed scheme is implemented with a system‐onchip of fully integrated analog CIM modules and digital sub‐systems, showing fast convergence of LeNet training to 97.73%. The efficacy of training larger models is evaluated using realistic hardware parameters and verifies that CIM modules can enable efficient mix‐precision DNN training with accuracy comparable to full‐precision software‐trained models. Additionally, models trained on chip are inherently robust to hardware variations, allowing direct mapping to CIM inference chips without additional re‐training.