An official website of the United States government
Deep learning models rely heavily on extensive training data, but obtaining sufficient real-world data remains a major challenge in clinical fields. To address this, we explore methods for generating realistic synthetic multivariate fall data to supplement limited real-world samples collected from three fall-related datasets: SmartFallMM, UniMib, and K-Fall. We apply three conventional time-series augmentation techniques, a Diffusion-based generative AI method, and a novel approach that extracts fall segments from public video footage of older adults. A key innovation of our work is the exploration of two distinct approaches: video-based pose estimation to extract fall segments from public footage, and Diffusion models to generate synthetic fall signals. Both methods independently enable the creation of highly realistic and diverse synthetic data tailored to specific sensor placements. To our knowledge, these approaches and especially their application in fall detection represent rarely explored directions in this research area. To assess the quality of the synthetic data, we use quantitative metrics, including the Fréchet Inception Distance (FID), Discriminative Score, Predictive Score, Jensen–Shannon Divergence (JSD), and Kolmogorov–Smirnov (KS) test, and visually inspect temporal patterns for structural realism. We observe that Diffusion-based synthesis produces the most realistic and distributionally aligned fall data. To further evaluate the impact of synthetic data, we train a long short-term memory (LSTM) model offline and test it in real time using the SmartFall App. Incorporating Diffusion-based synthetic data improves the offline F1-score by 7–10% and boosts real-time fall detection performance by 24%, confirming its value in enhancing model robustness and applicability in real-world settings.
more »
« less
Combining Perception Considerations with Artificial Intelligence in Maritime Threat Detection Systems
Over the past few years the need for early-warning maritime threat detection systems has dramatically increased. Our research aims to address this need by tackling three main problems: 1) classify boat activities into three categories: random walk, following, and chasing, 2) real-time classification of boat path trajectories, and 3) designing a novel perception-based framework for activity detection in the maritime context. We propose the implementation of an entropy-based detection algorithm, trained using synthetic data. We assess the viability of the proposed framework based on accuracy and the number of time steps required prior to identification. The synthetic data generated has the potential to spur other research efforts in the field of maritime detection.
more »
« less
- PAR ID:
- 10348095
- Date Published:
- Journal Name:
- 2022 17th Annual System of Systems Engineering Conference (SOSE)
- Page Range / eLocation ID:
- 417 to 422
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Network-on-Chip (NoC) is widely employed by multi-core System-on-Chip (SoC) architectures to cater to their communication requirements. Increasing NoC complexity coupled with its widespread usage has made it a focal point of potential security attacks. Distributed Denial-of-Service (DDoS) is one such attack that is caused by malicious intellectual property (IP) cores flooding the network with unnecessary packets causing significant performance degradation through NoC congestion. In this paper, we propose an efficient framework for real-time detection and localization of DDoS attacks. This paper makes three important contributions. We propose a real-time and lightweight DDoS attack detection technique for NoC-based SoCs by monitoring packets to detect any violations. Once a potential attack has been flagged, our approach is also capable of localizing the malicious IPs using the latency data in the NoC routers. The applications are statically profiled during design time to determine communication patterns. These patterns are then used for real-time detection and localization of DDoS attacks. We have evaluated the effectiveness of our approach against different NoC topologies and architecture models using both real benchmarks and synthetic traffic patterns. Our experimental results demonstrate that our proposed approach is capable of real-time detection and localization of DDoS attacks originating from multiple malicious IPs in NoC-based SoCs.more » « less
-
Cyber-physical systems (CPS) have been increasingly attacked by hackers. CPS are especially vulnerable to attackers that have full knowledge of the system's configuration. Therefore, novel anomaly detection algorithms in the presence of a knowledgeable adversary need to be developed. However, this research is still in its infancy due to limited attack data availability and test beds. By proposing a holistic attack modeling framework, we aim to show the vulnerability of existing detection algorithms and provide a basis for novel sensor-based cyber-attack detection. Stealthy Attack GEneration (SAGE) for CPS serves as a tool for cyber-risk assessment of existing systems and detection algorithms for practitioners and researchers alike. Stealthy attacks are characterized by malicious injections into the CPS through input, output, or both, which produce bounded changes in the detection residue. By using the SAGE framework, we generate stealthy attacks to achieve three objectives: (i) Maximize damage, (ii) Avoid detection, and (iii) Minimize the attack cost. Additionally, an attacker needs to adhere to the physical principles in a CPS (objective iv). The goal of SAGE is to model worst-case attacks, where we assume limited information asymmetries between attackers and defenders (e.g., insider knowledge of the attacker). Those worst-case attacks are the hardest to detect, but common in practice and allow understanding of the maximum conceivable damage. We propose an efficient solution procedure for the novel SAGE optimization problem. The SAGE framework is illustrated in three case studies. Those case studies serve as modeling guidelines for the development of novel attack detection algorithms and comprehensive cyber-physical risk assessment of CPS. The results show that SAGE attacks can cause severe damage to a CPS, while only changing the input control signals minimally. This avoids detection and keeps the cost of an attack low. This highlights the need for more advanced detection algorithms and novel research in cyber-physical security.more » « less
-
Abstract Synthetic ensemble forecasts are an important tool for testing the robustness of forecast‐informed reservoir operations (FIRO). These forecasts are statistically generated to mimic the skill of hindcasts derived from operational ensemble forecasting systems, but they can be created for time periods when hindcast data are unavailable, allowing for a more comprehensive evaluation of FIRO policies. Nevertheless, it remains unclear how to determine whether a candidate synthetic ensemble forecasting approach is sufficiently representative of its real‐world counterpart to support FIRO policy evaluation. This highlights a need for formalfit‐for‐purposevalidation frameworks to advance synthetic forecasting as a generalizable risk analysis strategy. We address this research gap by first introducing a novel operations‐based validation framework, where reservoir storage and release simulations under a FIRO policy are compared when forced with a single ensemble hindcast and many different synthetic ensembles. We evaluate the suitability of synthetic forecasts based on formal probabilistic verification of the operational outcomes. Second, we develop a new synthetic ensemble forecasting algorithm and compare it to a previous algorithm using this validation framework across a set of stylized, hydrologically diverse reservoir systems in California. Results reveal clear differences in operational suitability, with the new method consistently outperforming the previous one. These findings demonstrate the promise of the newer synthetic forecasting approach as a generalizable tool for FIRO policy evaluation and robustness testing. They also underscore the value of the proposed validation framework in benchmarking and guiding future improvements in synthetic forecast development.more » « less
-
We present a framework for learning useful subgoals that support efficient long-term planning to achieve novel goals. At the core of our framework is a collection of rational subgoals (RSGs), which are essentially binary classifiers over the environmental states. RSGs can be learned from weakly-annotated data, in the form of unsegmented demonstration trajectories, paired with abstract task descriptions, which are composed of terms initially unknown to the agent (e.g., collect-wood then craft-boat then go-across-river). Our framework also discovers dependencies between RSGs, e.g., the task collect-wood is a helpful subgoal for the task craft-boat. Given a goal description, the learned subgoals and the derived dependencies facilitate off-the-shelf planning algorithms, such as A* and RRT, by setting helpful subgoals as waypoints to the planner, which significantly improves performance-time efficiency.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.1109/SOSE55472.2022.9812640
