Nous utilisons des témoins pour analyser le trafic et l’utilisation de notre site web, afin de personnaliser votre expérience. Vous pouvez désactiver ces technologies à tout moment, mais cela peut restreindre certaines fonctionnalités du site. Consultez notre Politique de protection de la vie privée pour en savoir plus.
Paramètre des cookies
Vous pouvez activer et désactiver les types de cookies que vous souhaitez accepter. Cependant certains choix que vous ferez pourraient affecter les services proposés sur nos sites (ex : suggestions, annonces personnalisées, etc.).
Cookies essentiels
Ces cookies sont nécessaires au fonctionnement du site et ne peuvent être désactivés. (Toujours actif)
Cookies analyse
Acceptez-vous l'utilisation de cookies pour mesurer l'audience de nos sites ?
Multimedia Player
Acceptez-vous l'utilisation de cookies pour afficher et vous permettre de regarder les contenus vidéo hébergés par nos partenaires (YouTube, etc.) ?
Publications
Development of AI-assisted microscopy frameworks through realistic simulation with pySTED
The Bayesian approach leads to coherent updates of predictions under new data, which makes adhering to Bayesian principles appealing in deci… (voir plus)sion-making contexts. Traditionally, integrating Bayesian principles into models like deep neural networks involves setting priors on parameters and approximating posteriors. This is done despite the fact that, typically, priors on parameters reflect any prior beliefs only insofar as they dictate function space behaviour. In this paper, we rethink this approach and consider what properties characterise a prediction rule as being Bayesian. Algorithms meeting such criteria can be deemed implicitly Bayesian — they make the same predictions as some Bayesian model, without explicitly manifesting priors and posteriors. We argue this might be a more fruitful approach towards integrating Bayesian principles into deep learning. In this paper, we propose how to measure how close a general prediction rule is to being implicitly Bayesian, and empirically evaluate multiple prediction strategies using our approach. We also show theoretically that agents relying on non-implicitly Bayesian prediction rules can be easily exploited in adversarial betting settings.
2024-07-29
Proceedings of the 6th Symposium on Advances in Approximate Bayesian Inference (publié)
Single-cell multi-omics illuminate intricate cellular states, yielding transformative insights into cellular dynamics and disease. Yet, whil… (voir plus)e the potential of this technology is vast, the integration of its multifaceted data presents challenges. Some modalities have not reached the robustness or clarity of established scRNA-seq. Coupled with data scarcity for newer modalities and integration intricacies, these challenges limit our ability to maximize single-cell omics benefits. We introduce scCross: a tool adeptly engineered using variational autoencoder, generative adversarial network principles, and the Mutual Nearest Neighbors (MNN) technique for modality alignment. This synergy ensures seamless integration of varied single-cell multi-omics data. Beyond its foundational prowess in multi-omics data integration, scCross excels in single-cell cross-modal data generation, multi-omics data simulation, and profound in-silico cellular perturbations. Armed with these capabilities, scCross is set to transform the field of single-cell research, establishing itself in the nuanced integration, generation, and simulation of complex multi-omics data.
Incident reporting and learning systems provide an opportunity to identify systemic vulnerabilities that contribute to incidents and potenti… (voir plus)ally degrade quality. The narrative of an incident is intended to provide a clear, easy to understand description of an incident. Unclear, incomplete or poorly organized narratives compromise the ability to learn from them. This report provides guidance for drafting effective narratives, with particular attention to the use of narratives in incident reporting and learning systems (IRLS). Examples are given that compare effective and less than effective narratives. This report is mostly directed to organizations that maintain IRLS, but also may be helpful for individuals who desire to write a useful narrative for entry into such a system. Recommendations include the following: (1) Systems should allow a one- or two-sentence, free-text synopsis of an incident without guessing at causes; (2) Information included should form a sequence of events with chronology; and (3) Reporting and learning systems should consider using the headings suggested to guide the reporter through the narrative: (a) incident occurrences and actions by role; (b) prior circumstances and actions; (c) method by which the incident was identified; (d) equipment related details if relevant; (e) recovery actions by role; (f) relevant time span between responses; (g) and how individuals affected during or immediately after incident. When possible and appropriate, supplementary information including relevant data elements should be included using numerical scales or drop-down choices outside of the narrative. Information that should not be included in the narrative includes: (a) patient health information (PHI); (b) conjecture or blame; (c) jargon abbreviations or details without specifying their significance; (d) causal analysis.
Offline reinforcement learning has shown promise for solving tasks in safety-critical settings, such as clinical decision support. Its appli… (voir plus)cation, however, has been limited by the lack of interpretability and interactivity for clinicians. To address these challenges, we propose the medical decision transformer (MeDT), a novel and versatile framework based on the goal-conditioned reinforcement learning paradigm for sepsis treatment recommendation. MeDT uses the decision transformer architecture to learn a policy for drug dosage recommendation. During offline training, MeDT utilizes collected treatment trajectories to predict administered treatments for each time step, incorporating known treatment outcomes, target acuity scores, past treatment decisions, and current and past medical states. This analysis enables MeDT to capture complex dependencies among a patient's medical history, treatment decisions, outcomes, and short-term effects on stability. Our proposed conditioning uses acuity scores to address sparse reward issues and to facilitate clinician-model interactions, enhancing decision-making. Following training, MeDT can generate tailored treatment recommendations by conditioning on the desired positive outcome (survival) and user-specified short-term stability improvements. We carry out rigorous experiments on data from the MIMIC-III dataset and use off-policy evaluation to demonstrate that MeDT recommends interventions that outperform or are competitive with existing offline reinforcement learning methods while enabling a more interpretable, personalized and clinician-directed approach.
Reinforcement learning practitioners often avoid hierarchical policies, especially in image-based observation spaces. Typically, the single-… (voir plus)task performance improvement over flat-policy counterparts does not justify the additional complexity associated with implementing a hierarchy. However, by introducing multiple decision-making levels, hierarchical policies can compose lower-level policies to more effectively generalize between tasks, highlighting the need for multi-task evaluations. We analyze the benefits of hierarchy through simulated multi-task robotic control experiments from pixels. Our results show that hierarchical policies trained with task conditioning can (1) increase performance on training tasks, (2) lead to improved reward and state-space generalizations in similar tasks, and (3) decrease the complexity of fine tuning required to solve novel tasks. Thus, we believe that hierarchical policies should be considered when building reinforcement learning architectures capable of generalizing between tasks.
SETTING
Mathematical modelling played an important role in the public health response to COVID-19 in Canada. Variability in epidemic traject… (voir plus)ories, modelling approaches, and data infrastructure across provinces provides a unique opportunity to understand the factors that shaped modelling strategies.
INTERVENTION
Provinces implemented stringent pandemic interventions to mitigate SARS-CoV-2 transmission, considering evidence from epidemic models. This study aimed to summarize provincial COVID-19 modelling efforts. We identified modelling teams working with provincial decision-makers, through referrals and membership in Canadian modelling networks. Information on models, data sources, and knowledge translation were abstracted using standardized instruments.
OUTCOMES
We obtained information from six provinces. For provinces with sustained community transmission, initial modelling efforts focused on projecting epidemic trajectories and healthcare demands, and evaluating impacts of proposed interventions. In provinces with low community transmission, models emphasized quantifying importation risks. Most of the models were compartmental and deterministic, with projection horizons of a few weeks. Models were updated regularly or replaced by new ones, adapting to changing local epidemic dynamics, pathogen characteristics, vaccines, and requests from public health. Surveillance datasets for cases, hospitalizations and deaths, and serological studies were the main data sources for model calibration. Access to data for modelling and the structure for knowledge translation differed markedly between provinces.
IMPLICATION
Provincial modelling efforts during the COVID-19 pandemic were tailored to local contexts and modulated by available resources. Strengthening Canadian modelling capacity, developing and sustaining collaborations between modellers and governments, and ensuring earlier access to linked and timely surveillance data could help improve pandemic preparedness.
Language models exhibit scaling laws, whereby increasing model and dataset size yields predictable decreases in negative log likelihood, unl… (voir plus)ocking a dazzling array of capabilities. This phenomenon spurs many companies to train ever larger models in pursuit of ever improved performance. Yet, these models are vulnerable to adversarial inputs such as ``jailbreaks'' and prompt injections that induce models to perform undesired behaviors, posing a growing risk as models become more capable. Prior work indicates that computer vision models become more robust with model and data scaling, raising the question: does language model robustness also improve with scale? We study this question empirically in the classification setting, finding that without explicit defense training, larger models tend to be modestly more robust on most tasks, though the effect is not reliable. Even with the advantage conferred by scale, undefended models remain easy to attack in absolute terms, and we thus turn our attention to explicitly training models for adversarial robustness, which we show to be a much more compute-efficient defense than scaling model size alone. In this setting, we also observe that adversarially trained larger models generalize faster and better to modified attacks not seen during training when compared with smaller models. Finally, we analyze the offense/defense balance of increasing compute, finding parity in some settings and an advantage for offense in others, suggesting that adversarial training alone is not sufficient to solve robustness, even at greater model scales.
