help研究问题HAR模型鲁棒性、泛化与数据增强
混合数据增强技术（如Mixup）如何在跨域场景中避免语义不一致和分类边界模糊问题？UbiComp '22Semantic-Discriminative Mixup for Generalizable Sensor-based Cross-domain Activity Recognition
所属方向: 传感、识别与安全
HAR模型鲁棒性、泛化与数据增强
统计基于当前已索引 question 数据，来源缺失不影响问题展示。
115
问题/痛点
30
来源
2024
最近年份
help研究问题HAR模型鲁棒性、泛化与数据增强
语义范围（Activity Semantic Range）和大间隔损失（Large Margin Loss）能否提升模型的分类鲁棒性和准确性？UbiComp '22Semantic-Discriminative Mixup for Generalizable Sensor-based Cross-domain Activity Recognition
lightbulb现实痛点HAR模型鲁棒性、泛化与数据增强
传感器活动识别模型在目标域数据不可访问时缺乏泛化能力。UbiComp '22Semantic-Discriminative Mixup for Generalizable Sensor-based Cross-domain Activity Recognition
help研究问题HAR模型鲁棒性、泛化与数据增强
如何利用IMU传感器的多模态数据提高人类活动识别的准确性？UbiComp '22IF-ConvTransformer: A Framework for Human Activity Recognition Using IMU Fusion and ConvTransformer
help研究问题HAR模型鲁棒性、泛化与数据增强
如何通过融合卷积网络和Transformer结构有效捕捉局部特征和全局时间依赖？UbiComp '22IF-ConvTransformer: A Framework for Human Activity Recognition Using IMU Fusion and ConvTransformer
help研究问题HAR模型鲁棒性、泛化与数据增强
如何解决传统RNN在捕捉长时间依赖关系时的“遗忘”问题？UbiComp '22IF-ConvTransformer: A Framework for Human Activity Recognition Using IMU Fusion and ConvTransformer
lightbulb现实痛点HAR模型鲁棒性、泛化与数据增强
可穿戴设备在识别人类复杂活动时准确率较低。UbiComp '22IF-ConvTransformer: A Framework for Human Activity Recognition Using IMU Fusion and ConvTransformer
help研究问题HAR模型鲁棒性、泛化与数据增强
多通道可穿戴传感器如何通过自注意力机制优化其活动识别表现？UbiComp '21Attend and Discriminate: Beyond the State-of-the-Art for Human Activity Recognition Using Wearable Sensors
help研究问题HAR模型鲁棒性、泛化与数据增强
如何在优化活动识别性能时兼顾类内紧凑性与类间可分离性？UbiComp '21Attend and Discriminate: Beyond the State-of-the-Art for Human Activity Recognition Using Wearable Sensors
help研究问题HAR模型鲁棒性、泛化与数据增强
如何通过数据增强方法生成保持活动标签语义完整的虚拟样本？UbiComp '21Attend and Discriminate: Beyond the State-of-the-Art for Human Activity Recognition Using Wearable Sensors
lightbulb现实痛点HAR模型鲁棒性、泛化与数据增强
可穿戴设备检测人体活动时，准确度和鲁棒性不足UbiComp '21Attend and Discriminate: Beyond the State-of-the-Art for Human Activity Recognition Using Wearable Sensors
help研究问题HAR模型鲁棒性、泛化与数据增强
如何使用视频嵌入构建语义空间，以支持IMU数据的人体活动无监督分类？UbiComp '21Zero-Shot Learning for IMU-Based Activity Recognition Using Video Embeddings
help研究问题HAR模型鲁棒性、泛化与数据增强
跨模态知识转移能否有效提高IMU数据中未见活动类别的识别准确性？UbiComp '21Zero-Shot Learning for IMU-Based Activity Recognition Using Video Embeddings
help研究问题HAR模型鲁棒性、泛化与数据增强
将视频嵌入与文本嵌入相结合是否能进一步优化IMU活动识别性能？UbiComp '21Zero-Shot Learning for IMU-Based Activity Recognition Using Video Embeddings
lightbulb现实痛点HAR模型鲁棒性、泛化与数据增强
现有IMU系统无法识别新活动类别，影响日常应用如健康监测。UbiComp '21Zero-Shot Learning for IMU-Based Activity Recognition Using Video Embeddings