Model interpretability plays a central role in human-AI decision-making systems. Ideally, explanations should be expressed through semantic concepts and their causal relations in an interpretable way for the human experts. Additionally, explanation methods should be efficient, and not compromise the … Read More
Author Archives: admin
Scaling Laws for Multilingual Neural Machine Translation
In this talk, we provide a large-scale empirical study of the scaling properties of multilingual neural machine translation models. We examine how increases in the model size affect the model performance and investigate the role of the training mixture composition … Read More
Artificial evolution in the natural world
Evolutionary algorithms (EA) have proved as a robust optimisation meta-heuristic in engineering problems. When we tackle problems in multi-agent systems, the nonlinear interactions set up an increasingly challenging context. And if the multi-agents are groups of animals interacting with artificial … Read More
Model-Value Self-Consistent Updates and Applications
Learned models of the environment provide reinforcement learning agents with flexible ways of making predictions about the environment. Models enable planning, i.e. using more computation to improve value functions or policies, without requiring additional environment interactions. In this talk, we … Read More
FairGBM: Gradient Boosting with Fairness Constraints
Tabular data is prevalent in many high-stakes domains, from financial services to public policy. In these settings, Gradient Boosted Machines (GBM) are still the state-of-the-art. However, existing in-training fairness interventions are either incompatible with GBMs, or incur significant performance losses … Read More
Dense Template Retrieval for Customer Support
Templated answers are used extensively in customer support scenarios, providing an efficient way to cover a plethora of topics, with an easily maintainable collection of templates. However, the number of templates is often too high for an agent to manually … Read More
ML-Inception: understanding where and why models work (and don’t work)
A subgroup discovery-based method has recently been proposed to understand the behavior of models in the (original) feature space. The subgroups identified represent areas of feature space where the model obtains better or worse predictive performance than on average. For … Read More
Machine Learning for Motion Planning of Autonomous Vehicles — Interaction-Aware Motion Planning in Crowded Dynamic Environments
Robotic navigation in environments shared with other robots or humans remains challenging as the intentions of the surrounding agents are not directly observable. Moreover, interaction is crucial to enable safe and efficient navigation in crowded scenarios. Local trajectory optimization methods, … Read More
On the Calibration of Generative Question-Answering models: State-of-the-art and Challenges
Nowadays, generative question-answering models (e.g., UnifiedQA) achieve state-of-the-art performance in various datasets. Despite their remarkable performance, these models still produce wrong answers with high confidence scores. The responsible use of such systems in high-risk applications, like healthcare, requires some guarantees … Read More
Explainable Artificial Intelligence and its role in supporting medical diagnosis
The use of deep learning algorithms in the clinical context is hindered by their lack of interpretability. One way of increasing the acceptance of such complex algorithms is by providing explanations of the decisions through the presentation of similar examples. … Read More