This study proposes a novel flexible hybrid state estimation (SE) algorithm when a realistic communication system with its irregularities is taken into account. This system is modelled by the Network Simulator 2 software tool, which is also used to calculate communication delays and packet drop probabilities. Within this setup, the system observability can be predicted, and the proposed SE can decide between using the static SE (SSE) or the discrete Kalman filter plus SSE‐based measurements and time alignment (Forecasting‐aided SE). Flexible hybrid SE (FHSE) incorporates both phasor measurement units and supervisory control and data acquisition‐based measurements, with different time stamps. The proposed FHSE with detailed modelling of the communication system is motivated by: (i) well‐known issues in SSE (time alignment of the measurements, frequent un‐observability for fixed SE time stamps etc.); and (ii) the need to model a realistic communication system (calculated communication delays and packet drop probabilities are a part of the proposed FHSE). Application of the proposed algorithm is illustrated for examples with time‐varying bus load/generation on two IEEE test cases: 14‐bus and 300‐bus.

This paper presents a procedure for estimating the systems state when considerable Information and Communication Technology (ICT) component outages occur, leaving entire system areas unobservable. For this task, a novel method for analyzing system observability is proposed based on the Manifold Boundary Approximation Method (MBAM). By utilizing information geometry, MBAM analyzes boundaries of models in data space, thus detecting unidentifiable system parameters and states based on available data. This approach extends local, matrix-based methods to a global perspective, making it capable of detecting both structurally unidentifiable parameters as well as practically unidentifiable parameters (i.e., identifiable with low accuracy). Beyond partitioning identifiable/unidentifiable states, MBAM also reduces the model to remove reference to the unidentifiable state variables. To test this procedure, cyber-physical system (CPS) simulation environments are constructed by co-simulating the physical and cyber system layers.

Annotated medical imaging data remain scarce, and labels are often weak and noisy: in mammography, an examination comprises several high-resolution views, yet the diagnostic outcome is recorded only at the breast level. Such problems are naturally cast as Multiple Instance Learning (MIL), where the model must infer instance-level structure from bag-level labels alone. Although contemporary foundation encoders supply strong general-purpose embeddings, augmenting MIL data in this representation space remains an open problem as established techniques act on one instance at a time and ignore the statistical dependencies binding a bag together. We address this with SetFlow, a generative model that learns the distribution of complete MIL bags directly in a frozen encoder’s embedding space. SetFlow couples flow-matching training with a Set Transformer-inspired backbone, making it invariant to instance ordering while modeling intra-bag relationships. Generation is conditioned jointly on class label and per-instance scale, yielding coherent, semantically faithful bags rather than isolated vectors. Evaluating on two large public mammography datasets and two encoders, we assess distributional fidelity, nearest-neighbor behavior, and downstream augmentation utility. We show that generated bags reproduce real-data statistics and improve classification in certain configuration, with performance gains varying on the amount of synthetic data. An architecture ablation confirms each design choice contributes to performance.