RESUMEN

Intuitively, experience playing against one mixture of opponents in a given domain should be relevant for a different mixture in the same domain. If the mixture changes, ideally we would not have to train from scratch, but rather could transfer what we have learned to construct a policy to play against the new mixture. We propose a transfer learning method, Q-Mixing, that starts by learning Q-values against each pure-strategy opponent. Then a Q-value for any distribution of opponent strategies is approximated by appropriately averaging the separately learned Q-values. From these components, we construct policies against all opponent mixtures without any further training. We empirically validate Q-Mixing in two environments: a simple grid-world soccer environment, and a social dilemma game. Our experiments find that Q-Mixing can successfully transfer knowledge across any mixture of opponents. Next, we consider the use of observations during play to update the believed distribution of opponents. We introduce an opponent policy classifier-trained reusing Q-learning data-and use the classifier results to refine the mixing of Q-values. Q-Mixing augmented with the opponent policy classifier performs better, with higher variance, than training directly against a mixed-strategy opponent.

RESUMEN

Highly mutable infectious disease pathogens (hm-IDPs) such as HIV and influenza evolve faster than the human immune system can contain them, allowing them to circumvent traditional vaccination approaches and causing over one million deaths annually. Agent-based models can be used to simulate the complex interactions that occur between immune cells and hm-IDP-like proteins (antigens) during affinity maturation-the process by which antibodies evolve. Compared to existing experimental approaches, agent-based models offer a safe, low-cost, and rapid route to study the immune response to vaccines spanning a wide range of design variables. However, the highly stochastic nature of affinity maturation and vast sequence space of hm-IDPs render brute force searches intractable for exploring all pertinent vaccine design variables and the subset of immunization protocols encompassed therein. To address this challenge, we employed deep reinforcement learning to drive a recently developed agent-based model of affinity maturation to focus sampling on immunization protocols with greater potential to improve the chosen metrics of protection, namely the broadly neutralizing antibody (bnAb) titers or fraction of bnAbs produced. Using this approach, we were able to coarse-grain a wide range of vaccine design variables and explore the relevant design space. Our work offers new testable insights into how vaccines should be formulated to maximize protective immune responses to hm-IDPs and how they can be minimally tailored to account for major sources of heterogeneity in human immune responses and various socioeconomic factors. Our results indicate that the first 3 to 5 immunizations, depending on the metric of protection, should be specially tailored to achieve a robust protective immune response, but that beyond this point further immunizations require only subtle changes in formulation to sustain a durable bnAb response.

Asunto(s)

Vacunas contra el SIDA , Infecciones por VIH , VIH-1 , Humanos , Anticuerpos Anti-VIH , Anticuerpos Neutralizantes , Anticuerpos ampliamente neutralizantes , Infecciones por VIH/prevención & control

RESUMEN

Adversarial attacks, e.g., adversarial perturbations of the input and adversarial samples, pose significant challenges to machine learning and deep learning techniques, including interactive recommendation systems. The latent embedding space of those techniques makes adversarial attacks challenging to detect at an early stage. Recent advance in causality shows that counterfactual can also be considered one of the ways to generate the adversarial samples drawn from different distribution as the training samples. We propose to explore adversarial examples and attack agnostic detection on reinforcement learning (RL)-based interactive recommendation systems. We first craft different types of adversarial examples by adding perturbations to the input and intervening on the casual factors. Then, we augment recommendation systems by detecting potential attacks with a deep learning-based classifier based on the crafted data. Finally, we study the attack strength and frequency of adversarial examples and evaluate our model on standard datasets with multiple crafting methods. Our extensive experiments show that most adversarial attacks are effective, and both attack strength and attack frequency impact the attack performance. The strategically-timed attack achieves comparative attack performance with only 1/3 to 1/2 attack frequency. Besides, our white-box detector trained with one crafting method has the generalization ability over several other crafting methods.

RESUMEN

Deep reinforcement learning (DRL) combines reinforcement learning algorithms with deep neural networks (DNNs). Spiking neural networks (SNNs) have been shown to be a biologically plausible and energy efficient alternative to DNNs. Since the introduction of surrogate gradient approaches that allowed to overcome the discontinuity in the spike function, SNNs can now be trained with the backpropagation through time (BPTT) algorithm. While largely explored on supervised learning problems, little work has been done on investigating the use of SNNs as function approximators in DRL. Here we show how SNNs can be applied to different DRL algorithms like Deep Q-Network (DQN) and Twin-Delayed Deep Deteministic Policy Gradient (TD3) for discrete and continuous action space environments, respectively. We found that SNNs are sensitive to the additional hyperparameters introduced by spiking neuron models like current and voltage decay factors, firing thresholds, and that extensive hyperparameter tuning is inevitable. However, we show that increasing the simulation time of SNNs, as well as applying a two-neuron encoding to the input observations helps reduce the sensitivity to the membrane parameters. Furthermore, we show that randomizing the membrane parameters, instead of selecting uniform values for all neurons, has stabilizing effects on the training. We conclude that SNNs can be utilized for learning complex continuous control problems with state-of-the-art DRL algorithms. While the training complexity increases, the resulting SNNs can be directly executed on neuromorphic processors and potentially benefit from their high energy efficiency.