1,721,598 research outputs found
Local dimension reduction of summary statistics for likelihood-free inference
No full textApproximate Bayesian computation (ABC) and other likelihood-free inference methods have gained popularity in the last decade, as they allow rigorous statistical inference for complex models without analytically tractable likelihood functions. A key component for accurate inference with ABC is the choice of summary statistics, which summarize the information in the data, but at the same time should be low-dimensional for efficiency. Several dimension reduction techniques have been introduced to automatically construct informative and low-dimensional summaries from a possibly large pool of candidate summaries. Projection-based methods, which are based on learning simple functional relationships from the summaries to parameters, are widely used and usually perform well, but might fail when the assumptions behind the transformation are not satisfied. We introduce a localization strategy for any projection-based dimension reduction method, in which the transformation is estimated in the neighborhood of the observed data instead of the whole space. Localization strategies have been suggested before, but the performance of the transformed summaries outside the local neighborhood has not been guaranteed. In our localization approach the transformation is validated and optimized over validation datasets, ensuring reliable performance. We demonstrate the improvement in the estimation accuracy for localized versions of linear regression and partial least squares, for three different models of varying complexity.Peer reviewe
Inference of Strategic Behavior based on Incomplete Observation Data
No full textInferring the goals, preferences and restrictions of strategically behaving agents is a common goal in many situations, and an important requirement for enabling computer systems to better model and understand human users. Inverse reinforcement learning (IRL) is one method for performing this kind of inference based on observations of the agent's behavior. However, traditional IRL methods are only applicable when the observations are in the form of state-action paths -- an assumption which does not hold in many real-world modelling settings. This paper demonstrates that inference is possible even with an arbitrary observation noise model
Modelling Human Decision-making based on Aggregate Observation Data
No full textBeing able to infer the goals, preferences and limitations of humans is of key importance in designing interactive systems. Reinforcement learning (RL) models are a promising direction of research, as they are able to model how the behavioural patterns of users emerge from the task and environment structure. One limitation with traditional inference methods for RL models is the strict requirements for observation data; both the states of the environment and the actions of the agent need to be observed at each step of the task. This has prevented RL models from being used in situations where such fine-grained observations are not available. In this extended abstract we present results from a recent study where we demonstrated how inference can be performed for RL models even when the observation data is significantly more coarse-grained. The idea is to solve the inverse reinforcement learning (IRL) problem using approximate Bayesian computation sped up with Bayesian optimization
Deep Bayesian Experimental Design for Drug Discovery
No full textPublisher Copyright: © The Author(s) 2025. | openaire: EC/H2020/956832/EU//AIDDIn drug discovery, prioritizing compounds for testing is an important task. Active learning can assist in this endeavor by prioritizing molecules for label acquisition based on their estimated potential to enhance in-silico models. However, in specialized cases like toxicity modeling, limited dataset sizes can hinder effective training of modern neural networks for representation learning and to perform active learning. In this study, we leverage a transformer-based BERT model pretrained on millions of SMILES to perform active learning. Additionally, we explore different acquisition functions to assess their compatibility with pretrained BERT model. Our results demonstrate that pretrained models enhance active learning outcomes. Furthermore, we observe that active learning selects a higher proportion of positive compounds compared to random acquisition functions, an important advantage, especially in dealing with imbalanced toxicity datasets. Through a comparative analysis, we find that both BALD and EPIG acquisition functions outperform random acquisition, with EPIG exhibiting slightly superior performance over BALD. In summary, our study highlights the effectiveness of active learning in conjunction with pretrained models to tackle the problem of data scarcity.Peer reviewe
Approximate Bayesian Computation with Domain Expert in the Loop
No full textApproximate Bayesian computation (ABC) is a popular likelihood-free inference
method for models with intractable likelihood functions. As ABC methods usually
rely on comparing summary statistics of observed and simulated data, the choice
of the statistics is crucial. This choice involves a trade-off between loss of
information and dimensionality reduction, and is often determined based on
domain knowledge. However, handcrafting and selecting suitable statistics is a
laborious task involving multiple trial-and-error steps. In this work, we
introduce an active learning method for ABC statistics selection which reduces
the domain expert's work considerably. By involving the experts, we are able to
handle misspecified models, unlike the existing dimension reduction methods.
Moreover, empirical results show better posterior estimates than with existing
methods, when the simulation budget is limited.Comment: Accepted for publication at ICML 2022. Code available at
https://github.com/lfilstro/HITL-AB
Molecular property prediction using pretrained-BERT and Bayesian active learning : a data-efficient approach to drug design
No full textPublisher Copyright: © The Author(s) 2025. | openaire: EC/H2020/956832/EU//AIDDAbstract: In drug discovery, prioritizing compounds for experimental testing is a critical task that can be optimized through active learning by strategically selecting informative molecules. Active learning typically trains models on labeled examples alone, while unlabeled data is only used for acquisition. This fully supervised approach neglects valuable information present in unlabeled molecular data, impairing both predictive performance and the molecule selection process. We address this limitation by integrating a transformer-based BERT model, pretrained on 1.26 million compounds, into the active learning pipeline. This effectively disentangles representation learning and uncertainty estimation, leading to more reliable molecule selection. Experiments on Tox21 and ClinTox datasets demonstrate that our approach achieves equivalent toxic compound identification with 50% fewer iterations compared to conventional active learning. Analysis reveals that pretrained BERT representations generate a structured embedding space enabling reliable uncertainty estimation despite limited labeled data, confirmed through Expected Calibration Error measurements. This work establishes that combining pretrained molecular representations with active learning significantly improves both model performance and acquisition efficiency in drug discovery, providing a scalable framework for compound prioritization. Scientific Contribution: We demonstrate that high-quality molecular representations fundamentally determine active learning success in drug discovery, outweighing acquisition strategy selection. We provide a framework that integrates pretrained transformer models with Bayesian active learning to separate representation learning from uncertainty estimation—a critical distinction in low-data scenarios. This approach establishes a foundation for more efficient screening workflows across diverse pharmaceutical applications.Peer reviewe
Harmonizable mixture kernels with variational Fourier features
No full textThe expressive power of Gaussian processes depends heavily on the choice of kernel. In this work we propose the novel harmonizable mixture kernel (HMK), a family of expressive, interpretable, non-stationary kernels derived from mixture models on the generalized spectral representation. As a theoretically sound treatment of non-stationary kernels, HMK supports harmonizable covariances, a wide subset of kernels including all stationary and many non-stationary covariances. We also propose variational Fourier features, an inter-domain sparse GP inference framework that offers a representative set of 'inducing frequencies'. We show that harmonizable mixture kernels interpolate between local patterns, and that variational Fourier features offers a robust kernel learning framework for the new kernel family.Peer reviewe
Can relevance of images be inferred from eye movements?
No full textSearching for images from a large collection is a difficult task for automated algorithms. Many current techniques rely on items which have been manually 'tagged' with descriptors. This situation is not ideal, as it is difficult to formulate the initial query, and navigate the large number of hits returned. In order to present relevant images to the user, many systems rely on an explicit feedback mechanism. A machine learning algorithm can be used to present a new set of relevant images to the user -- thus increasing hit rates. In this work we use eye movements to assist a user when performing such a task, and ask this basic question: "Is it possible to replace or complement scarce explicit feedback with implicit feedback inferred from various sensors not specifically designed for the task?" We give initial results on a range of tasks and experiments which extend those presented in the Multimedia Information Retrieval conference (MIR'08). In reasonably controlled setups, fairly simple eye movements’ features in conjunction with machine learning techniques are capable of judging the relevance of an image based on eye movements alone, without using any explicit feedback -- therefore potentially assisting the user in a task
Learning Global Pairwise Interactions with Bayesian Neural Networks
No full textEstimating global pairwise interaction effects, i.e., the difference between the joint effect and the sum of marginal effects of two input features, with uncertainty properly quantified, is centrally important in science applications. We propose a non-parametric probabilistic method for detecting interaction effects of unknown form. First, the relationship between the features and the output is modelled using a Bayesian neural network, capable of representing complex interactions and principled uncertainty. Second, interaction effects and their uncertainty are estimated from the trained model. For the second step, we propose an intuitive global interaction measure: Bayesian Group Expected Hessian (GEH), which aggregates information of local interactions as captured by the Hessian. GEH provides a natural trade-off between type I and type II error and, moreover, comes with theoretical guarantees ensuring that the estimated interaction effects and their uncertainty can be improved by training a more accurate BNN. The method empirically outperforms available non-probabilistic alternatives on simulated and real-world data. Finally, we demonstrate its ability to detect interpretable interactions between higher-level features (at deeper layers of the neural network).Peer reviewe
Predicting relevance of parts of an image
No full textThis report studies the task of inferring which parts of an image are relevant for the user viewing the image. The relevance is inferred from gaze trajectory of users viewing the images given a specific task. Novel computational models based on both Bayesian generative modeling and kernel methods are developed for inferring the regions of interest from raw fixation data, as well as from combination of eye movements and image content features
- …
