Recent & Upcoming Talks

Sylwia will talk about her STEM career starting from choosing technical studies and ending at her current positon at AstraZeneca. Her speech will cover her academic and professional career, together with volonteering activities - participating in open-source AI4Good projects, and serving as a mentor for kids.

Sylwia will talk about advantages and limitations of sign language corpora for sign language recognition via deep learning-based methods. Her speech will cover the challenges posed by the diversity of sign languages across nationalities and insights from the research conducted by the non-profit HearAI project.

We introduce the Adaptive Locked Agnostic Network (ALAN), a concept involving self-supervised visual feature extraction using a large backbone model to produce anatomically robust semantic self-segmentation. In the ALAN methodology, this self-supervised training occurs only once on a large and diverse dataset. We applied the ALAN approach to three publicly available echocardiography datasets and designed two downstream models, one for segmenting a target anatomical region, and a second for echocardiogram view classification.

This research delves into the application of unconditional and conditional Generative Adversarial Networks (GANs) in both centralized and decentralized settings. The centralized approach replicates studies on a large but imbalanced skin lesion dataset, while the decentralized approach emulates a more realistic hospital scenario with data from three institutions. We meticulously assess the models’ performance in terms of fidelity, diversity, training speed, and the predictive capabilities of classifiers trained on synthetic data. Moreover, we delve into the explainability of the models, focusing on both global and local features. Crucially, we validate the authenticity of the generated samples by calculating the distance between real images and their respective projections in the latent space, addressing a key concern in such applications.

This work explores a usability of synthetic data for training visual deep learning models in real-world applications.

The Eye for AI program is uniquely designed to attract high-performing AI talent with a Master’s or Doctor’s degree and 0-3 years of work experience, typically in the fields of data management, data science and machine learning.

The detect-waste team conducted comprehensive research on Artificial Intelligence usage in waste detection and classification to fight the world’s waste pollution problem.

We evaluated models’ performance in terms of fidelity, diversity, speed of training, and predictive ability of classifiers trained on the generated synthetic data. In addition we provided explainability through exploration of latent space and embeddings projection focused both on global and local explanations.

We identified the gaps in current evaluation methodologies and proposed a more comprehensive and holistic evaluation framework for multi-modal motion prediction autonomus vehicle system.

We designed an automated tool to convert HamNoSys annotations into numerical labels for defined initial features of body and hand positions.

We explained the classifier’s diagnosis for skin cancer using both local and global techniques of explainable AI.

In the HearAI non-profit project, we investigated different multilingual open sign language corpora labeled by linguists in the language-agnostic HAMburg NOtation SYStem.

In our solution, we proposed computer-friendly numeric multilabels that greatly simplify the structure of the language-agnostic HamNoSys without significant loss of glos meaning.

In the HearAI non-profit project, we investigated different multilingual open sign language corpora labeled by linguists in the language-agnostic HAMburg NOtation SYStem.

Sahlgrenska Hospital, Region Halland and AI Sweden have decided to lead an effort of applying Federated Learning in practice. For this use case we made use of the SIIM-ISIC 2020 dataset and performed two main tasks. First, melanoma detection, and second, skin lesion image generation.

The use of healthcare data in the development of DL models is associated with challenges relating to personal data and regulatory issues. Patient data cannot be freely shared and is therefore limited in its usefulness for creating AI solutions. All of these issues can be addressed using different methods, showing the potential of artificial data for such use cases.

Neural networks can be applied to distinguish between melanoma and non-melanoma cases in a few seconds, which makes it a great help to the diagnosing physician.

We compared the performance of three well-known deep learning approaches for object detection on the AGAR dataset, namely two-stage, one-stage, and transformer-based neural networks.

We developed a new module within gprMax that can be used to simulate complex dispersive materials using multi-Debye expansions in an automatic manner. The module is capable of modelling Havriliak-Negami, Cole-Cole, Cole-Davidson, Jonscher, Complex-Refractive Index Models, and indeed any arbitrary dispersive material with real and imaginary permittivity specified by the user.

The detect-waste team conducted comprehensive research on Artificial Intelligence usage in waste detection and classification to fight the world’s waste pollution problem.

In this work, we present the results of numerical modelling of nonlinear pulse propagation in multimode optical fibers leading to discretized conical emission.

The field distribution of the same mode could be slightly different depending on wavelength and reflective index profile. To exhibit a dispersion characteristic for multimode fibers over a wide spectral range, a verification of refractive index value for a particular mode at a specific wavelength is needed.

Counting bacterial colonies is a fundamental task in microbiology. Currently, manual counting remains the gold standard. This is a timeconsuming and error-prone process, which requires a trained professional. To avoid these issues, the automated method can be applied for the task. The goal of our work was to design a model that counts and classifies bacterial colonies in Petri dishes using RGB images.

There is an urgent need of clarifying the potential and areas of applicability of numerical models to facilitate the accurate design of future nonlinear and multimode fiber devices. Therefore we performed the simulations of conical emission using two well-known numerical tools, namely, the unidirectional pulse propagation equation and the multimode generalized nonlinear Schrodinger equation.

Our simulations show that all-normal dispersion supercontinuum generated in optimized microstructured fiber (with hexagonal and kagome geometry) covers whole transparency window of silica glass.

We extended a theoretical model of self-organized SHG to include an attenuation and investigated the influence of fiber loss on the self-organized SHG process.