Extensions of the SHAP method: using Shapley value to interpret models with dependencies and survival analysis models SHAP (SHapley Additive exPlanations) is one of the most popular explanation methods. The widespread use is due to a solid theoretical basis from […]
“Use your data wisely” – Data-Centric NLP in the e-commerce domain Many recent NLP publications focus on increasing the model capacity and tuning the model’s architecture, “Data-Centric AI” suggests putting more effort into preparing high-quality data instead. In this talk, […]
Real-world Sequential Decision Making with High-dimensional, Noisy Observations In this talk, we present a general overview of our work on real-world sequential decision making. Specifically, we will focus on the area of sequential decision making with high-dimensional, noisy observations, with […]
Subgoal search Planning and search have been immensely studied topics in AI research. However, many questions still stay unanswered. In my talk, I will explore topics on the intersection of planning and learning. A general question is how learning methods […]
Multi-Agent Reinforcement Learning for Integrated Network of Adaptive Traffic Signal Controllers (MARLIN-ATSC) Population is steadily increasing worldwide, resulting in intractable traffic congestion in dense urban areas. Adaptive traffic signal control (ATSC) has shown strong potential to effectively alleviate urban traffic […]
Language models that write code In 2020 OpenAI has presented GPT-3 to the world: a transformer model that operated natural language and showcased high level of general skills and emergent meta-learning capabilities. Shortly thereafter, a research project got started that […]
Adapting Language Models for Conversational Tasks Improvements in pretraining large language models (LM) made training large-scale conversational encoders impractical. In this talk, we show that the transformation of LMs towards conversational encoders and adaptation of the task towards the original […]
A practical approach to measuring and using the prediction uncertainty of a natural language processing system Can a model with 90% accuracy be useful, if the task strictly requires 98%? The answer is, probably yes, but not always. Thanks to […]
Challenging Sens.AI: Towards CE marked deep learning software for brain tumor analysis We have witnessed the unprecedented success of deep learning in virtually all areas of science and industry, with medical image analysis not being an exception here. Although there […]
Into the unknown: how to minimize the risk of models making wrong decisions Machine learning models deployed on production often struggle to match the performance measured on the testing set that was extracted from the training data. The reason behind […]
