The pharmaceutical industry must adhere to rigorous regulations to meet specific quality standards. Additionally, the intricate nature of pharmaceutical manufacturing processes and long time to production necessitates timely detection of batch failures. AI/ML models are used for predictive maintenance in an automated and data-driven manner to detect these failures and aid timely intervention. However, these models require substantial amount of data for model training. This can lead to extended time-to-value before a predictive monitoring system can be deployed for any new process due to long process lead times. The current research proposes COSYNE, a generative AI-based approach to generate manufacturing digital twin, reducing the model development time by augmenting synthetic data with real data. The proposed solution is validated on a large pharmaceutical company’s batch manufacturing dataset, and the results are benchmarked across multiple dimensions of generation quality. Empirical results demonstrate that the proposed COSYNE outperforms the state-of-the-art approach by 2-3 times on average across all the generation quality metrics. Moreover, COSYNE enhances downstream AI/ML performance significantly through data augmentation and reduces time-to-value by creating high-fidelity digital twins with only 10% of real data and still achieve similar performance as current baseline trained on entire real data.

Visual saliency has been an increasingly active research area in the last ten years with dozens of saliency models recently published. Nowadays, one of the big challenges in the field is to find a way to fairly evaluate all of these models. In this paper, on human eye fixations, we compare the ranking of 12 state-of-the art saliency models using 12 similarity metrics. The comparison is done on Jian Li's database containing several hundreds of natural images. Based on Kendall concordance coefficient, it is shown that some of the metrics are strongly correlated leading to a redundancy in the performance metrics reported in the available benchmarks. On the other hand, other metrics provide a more diverse picture of models' overall performance. As a recommendation, three similarity metrics should be used to obtain a complete point of view of saliency model performance.

A recent study has shown that several gait-related artifact removal techniques are not helpful to improve the performances of an ultra-compact P300-based BCI system integrating only three electrodes and designed for out of the lab experiments. Moreover, the authors advise to use seven electrodes with a standard xDawn spatial filter - known to magnify the P300 response - in order to obtain the best performance/compactedness ratio. However, the positive impact of the xDawn filter and the gaitrelated artifact removal techniques on the BCI performance with seven EEG electrodes had not been precisely evaluated at that time. This is precisely what we do in this study using seven healthy subjects walking at three different speeds. Astonishingly, none of the methods, even the so-called xDawn spatial filter, does significantly outperform the raw data. Thereby, the recommendation considering a low-complexity four-state P300 BCI under ambulatory conditions would be to use filtered raw data without specific gait-related artifact removal techniques.