arXiv subjects

Rodrigo Carrasco-Davis

Publications and source records attributed to Rodrigo Carrasco-Davis.

6 recordsLinked to original sources

arXiv2026

Optimal Learning Rate Schedule for Balancing Effort and Performance

Learning how to learn efficiently is a fundamental challenge for biological agents and a growing concern for artificial ones. To learn effectively, an agent must regulate its learning speed, balancing the benefits of rapid improvement against the costs of effort, instability, or resource use. We introduce a normative framework that formalizes this problem as an optimal control process in which the agent maximizes cumulative performance while incurring a cost of learning. From this objective, we derive a closed-form solution for the optimal learning rate, which has the form of a closed-loop controller that depends only on the agent's current and expected future performance. Under mild assumptions, this solution generalizes across tasks and architectures and reproduces numerically optimized schedules in simulations. In simple learning models, we can mathematically analyze how agent and task parameters shape learning-rate scheduling as an open-loop control solution. Because the optimal policy depends on expectations of future performance, the framework predicts how overconfidence or underconfidence influence engagement and persistence, linking the control of learning speed to theories of self-regulated learning. We further show how a simple episodic memory mechanism can approximate the required performance expectations by recalling similar past learning experiences, providing a biologically plausible route to near-optimal behaviour. Together, these results provide a normative and biologically plausible account of learning speed control, linking self-regulated learning, effort allocation, and episodic memory estimation within a unified and tractable mathematical framework.

cs.LG↗

arXiv2025

Uncertainty Prioritized Experience Replay

Prioritized experience replay, which improves sample efficiency by selecting relevant transitions to update parameter estimates, is a crucial component of contemporary value-based deep reinforcement learning models. Typically, transitions are prioritized based on their temporal difference error. However, this approach is prone to favoring noisy transitions, even when the value estimation closely approximates the target mean. This phenomenon resembles the noisy TV problem postulated in the exploration literature, in which exploration-guided agents get stuck by mistaking noise for novelty. To mitigate the disruptive effects of noise in value estimation, we propose using epistemic uncertainty estimation to guide the prioritization of transitions from the replay buffer. Epistemic uncertainty quantifies the uncertainty that can be reduced by learning, hence reducing transitions sampled from the buffer generated by unpredictable random processes. We first illustrate the benefits of epistemic uncertainty prioritized replay in two tabular toy models: a simple multi-arm bandit task, and a noisy gridworld. Subsequently, we evaluate our prioritization scheme on the Atari suite, outperforming quantile regression deep Q-learning benchmarks; thus forging a path for the use of uncertainty prioritized replay in reinforcement learning agents.

cs.LG↗

arXiv2023

Meta-Learning Strategies through Value Maximization in Neural Networks

Biological and artificial learning agents face numerous choices about how to learn, ranging from hyperparameter selection to aspects of task distributions like curricula. Understanding how to make these meta-learning choices could offer normative accounts of cognitive control functions in biological learners and improve engineered systems. Yet optimal strategies remain challenging to compute in modern deep networks due to the complexity of optimizing through the entire learning process. Here we theoretically investigate optimal strategies in a tractable setting. We present a learning effort framework capable of efficiently optimizing control signals on a fully normative objective: discounted cumulative performance throughout learning. We obtain computational tractability by using average dynamical equations for gradient descent, available for simple neural network architectures. Our framework accommodates a range of meta-learning and automatic curriculum learning methods in a unified normative setting. We apply this framework to investigate the effect of approximations in common meta-learning algorithms; infer aspects of optimal curricula; and compute optimal neuronal resource allocation in a continual learning setting. Across settings, we find that control effort is most beneficial when applied to easier aspects of a task early in learning; followed by sustained effort on harder aspects. Overall, the learning effort framework provides a tractable theoretical test bed to study normative benefits of interventions in a variety of learning systems, as well as a formal account of optimal cognitive control strategies over learning trajectories posited by established theories in cognitive neuroscience.

cs.NE↗

arXiv2021

Alert Classification for the ALeRCE Broker System: The Real-time Stamp Classifier

We present a real-time stamp classifier of astronomical events for the ALeRCE (Automatic Learning for the Rapid Classification of Events) broker. The classifier is based on a convolutional neural network, trained on alerts ingested from the Zwicky Transient Facility (ZTF). Using only the \textit{science, reference} and \textit{difference} images of the first detection as inputs, along with the metadata of the alert as features, the classifier is able to correctly classify alerts from active galactic nuclei, supernovae (SNe), variable stars, asteroids and bogus classes, with high accuracy ($\sim$94\%) in a balanced test set. In order to find and analyze SN candidates selected by our classifier from the ZTF alert stream, we designed and deployed a visualization tool called SN Hunter, where relevant information about each possible SN is displayed for the experts to choose among candidates to report to the Transient Name Server database. From June 26th 2019 to February 28th 2021, we have reported 6846 SN candidates to date (11.8 candidates per day on average), of which 971 have been confirmed spectroscopically. Our ability to report objects using only a single detection means that 70\% of the reported SNe occurred within one day after the first detection. ALeRCE has only reported candidates not otherwise detected or selected by other groups, therefore adding new early transients to the bulk of objects available for early follow-up. Our work represents an important milestone toward rapid alert classifications with the next generation of large etendue telescopes, such as the Vera C. Rubin Observatory.

astro-ph.IM↗

arXiv2018

Deep Learning for Image Sequence Classification of Astronomical Events

We propose a new sequential classification model for astronomical objects based on a recurrent convolutional neural network (RCNN) which uses sequences of images as inputs. This approach avoids the computation of light curves or difference images. This is the first time that sequences of images are used directly for the classification of variable objects in astronomy. The second contribution of this work is the image simulation process. We generate synthetic image sequences that take into account the instrumental and observing conditions, obtaining a realistic, set of movies for each astronomical object. The simulated dataset is used to train our RCNN classifier. This approach allows us to generate datasets to train and test our RCNN model for different astronomical surveys and telescopes. We aim at building a simulated dataset whose distribution is close enough to the real dataset, so that a fine tuning could match the distributions between real and simulated dataset. To test the RCNN classifier trained with the synthetic dataset, we used real-world data from the High cadence Transient Survey (HiTS) obtaining an average recall of 85%, improved to 94% after performing fine tuning with 10 real samples per class. We compare the results of our model with those of a light curve random forest classifier. The proposed RCNN with fine tuning has a similar performance on the HiTS dataset compared to the light curve classifier, trained on an augmented training set with 10 real samples per class. The RCNN approach presents several advantages in an alert stream classification scenario, such as a reduction of the data pre-processing, faster online evaluation and easier performance improvement using a few real data samples. These results encourage us to use this method for alert brokers systems that will process alert streams generated by new telescopes such as the Large Synoptic Survey Telescope.

astro-ph.IM↗

arXiv2018

Enhanced Rotational Invariant Convolutional Neural Network for Supernovae Detection

In this paper, we propose an enhanced CNN model for detecting supernovae (SNe). This is done by applying a new method for obtaining rotational invariance that exploits cyclic symmetry. In addition, we use a visualization approach, the layer-wise relevance propagation (LRP) method, which allows finding the relevant pixels in each image that contribute to discriminate between SN candidates and artifacts. We introduce a measure to assess quantitatively the effect of the rotational invariant methods on the LRP relevance heatmaps. This allows comparing the proposed method, CAP, with the original Deep-HiTS model. The results show that the enhanced method presents an augmented capacity for achieving rotational invariance with respect to the original model. An ensemble of CAP models obtained the best results so far on the HiTS dataset, reaching an average accuracy of 99.53%. The improvement over Deep-HiTS is significant both statistically and in practice.

astro-ph.IM↗