Thinking agents for zero-shot generalization to qualitatively novel tasks The Obelisk Team Astera Institute Emeryville, USA Abstract Intelligent organisms can solve truly novel problems which they have never encountered before, either in their lifetime or their evolution. An important component of this capacity is the ability to "think", that is, to mentally manipulate objects, concepts and behaviors in order to plan and evaluate possible solutions to novel problems, even without environment interaction. To generate problems that are truly qualitatively novel, while still solvable zero-shot (by mental simulation), we use the combinatorial nature of environments: we train the agent while withholding a specific combination of the environment's elements. The novel test task, based on this combination, is thus guaranteed to be truly novel, while still mentally simulable since the agent has been exposed to each individual element (and their pairwise interactions) during training. We propose a method to train agents endowed with world models to make use their mental simulation abilities, by selecting tasks based on the difference between the agent's pre-thinking and post-thinking performance. When tested on the novel, withheld problem, the resulting agent successfully simulated alternative scenarios and used the resulting information to guide its behavior in the actual environment, solving the novel task in a single real-environment trial (zero-shot). 1 Introduction An important aspect of intelligence is the ability to handle novel problems. While simpler organisms are restricted to problems similar to these they have been exposed to during training, and fare badly when faced Correspondance: Thomas Miconi, thomas.miconi@gmail.comwith An major component of this capacity is the ability to think before acting. By'thinking' 1, that is, by internally manipulating concepts and behaviors and evaluating likely outcomes, agents can tackle novel problems never encountered before, by recombining existing knowledge into new solutions. This ability is perhaps the hallmark of what we think of as truly "intelligent" behavior: it is highly prevalent in humans, but is is debated whether it even exists in non-human animals [Suddendorf and Busby, 2003], including mammals such as rodents [Gillespie et al., 2021] or even great apes [Suddendorf et al., 2009, Os-vath, 2010]. Much work in machine learning has focused on training agents with increasingly complex innate behaviors.

Open-endedness stands to benefit from the ability to generate an infinite variety of diverse, challenging environments. One particularly interesting type of challenge is meta-learning ("learning-to-learn"), a hallmark of intelligent behavior. However, the number of meta-learning environments in the literature is limited. Here we describe a parametrized space for simple meta-reinforcement learning (meta-RL) tasks with arbitrary stimuli. The parametrization allows us to randomly generate an arbitrary number of novel simple meta-learning tasks. The parametrization is expressive enough to include many well-known meta-RL tasks, such as bandit problems, the Harlow task, T-mazes, the Daw two-step task and others. Simple extensions allow it to capture tasks based on two-dimensional topological spaces, such as full mazes or find-the-spot domains. We describe a number of randomly generated meta-RL domains of varying complexity and discuss potential issues arising from random generation.

Artificial neural networks (ANNs) have emerged as an essential tool in machine learning, achieving remarkable success across diverse domains, including image and speech generation, game playing, and robotics. However, there exist fundamental differences between ANNs' operating mechanisms and those of the biological brain, particularly concerning learning processes. This paper presents a comprehensive review of current brain-inspired learning representations in artificial neural networks. We investigate the integration of more biologically plausible mechanisms, such as synaptic plasticity, to enhance these networks' capabilities. Moreover, we delve into the potential advantages and challenges accompanying this approach. Ultimately, we pinpoint promising avenues for future research in this rapidly advancing field, which could bring us closer to understanding the essence of intelligence.

In one In meta-learning, networks are trained with external method, the "inner loop" stores information in the algorithms to learn tasks that require acquiring, time-varying activities of a recurrent network, which storing and exploiting unpredictable information for is slowly optimized in the "outer loop" over many each new instance of the task. However, animals are episodes [Hochreiter et al., 2001, Wang et al., 2016, able to pick up such cognitive tasks automatically, Duan et al., 2016]. A biological interpretation of as a result of their evolved neural architecture and this method is that the inner loop represents the synaptic plasticity mechanisms. Here we evolve neural within-episode self-sustaining activity of cerebral cortex, networks, endowed with plastic connections, over while the outer loop represents lifetime sculpting a sizeable set of simple meta-learning tasks based on of neural connections by value-based neural plasticity a framework from computational neuroscience. The (this interpretation is explored in detail by Wang resulting evolved network can automatically acquire et al. [2018]).

Continual learning is the problem of sequentially learning new tasks or knowledge while protecting previously acquired knowledge. However, catastrophic forgetting poses a grand challenge for neural networks performing such learning process. Thus, neural networks that are deployed in the real world often struggle in scenarios where the data distribution is non-stationary (concept drift), imbalanced, or not always fully available, i.e., rare edge cases. We propose a Differentiable Hebbian Consolidation model which is composed of a Differentiable Hebbian Plasticity (DHP) Softmax layer that adds a rapid learning plastic component (compressed episodic memory) to the fixed (slow changing) parameters of the softmax output layer; enabling learned representations to be retained for a longer timescale. We demonstrate the flexibility of our method by integrating well-known task-specific synaptic consolidation methods to penalize changes in the slow weights that are important for each target task. We evaluate our approach on the Permuted MNIST, Split MNIST and Vision Datasets Mixture benchmarks, and introduce an imbalanced variant of Permuted MNIST -- a dataset that combines the challenges of class imbalance and concept drift. Our proposed model requires no additional hyperparameters and outperforms comparable baselines by reducing forgetting.

Continual lifelong learning requires an agent or model to learn many sequentially ordered tasks, building on previous knowledge without catastrophically forgetting it. Much work has gone towards preventing the default tendency of machine learning models to catastrophically forget, yet virtually all such work involves manually-designed solutions to the problem. We instead advocate meta-learning a solution to catastrophic forgetting, allowing AI to learn to continually learn. Inspired by neuromodulatory processes in the brain, we propose A Neuromodulated Meta-Learning Algorithm (ANML). It differentiates through a sequential learning process to meta-learn an activation-gating function that enables context-dependent selective activation within a deep neural network. Specifically, a neuromodulatory (NM) neural network gates the forward pass of another (otherwise normal) neural network called the prediction learning network (PLN). The NM network also thus indirectly controls selective plasticity (i.e. the backward pass of) the PLN. ANML enables continual learning without catastrophic forgetting at scale: it produces state-of-the-art continual learning performance, sequentially learning as many as 600 classes (over 9,000 SGD updates).

A BSTRACT Standard gradient descent methods are susceptible to a range of issues that can impede training, such as high correlations and different scaling in parameter space. These difficulties can be addressed by second-order approaches that apply a preconditioning matrix to the gradient to improve convergence. Unfortunately, such algorithms typically struggle to scale to high-dimensional problems, in part because the calculation of specific preconditioners such as the inverse Hessian or Fisher information matrix is highly expensive. We introduce first-order preconditioning (FOP), a fast, scalable approach that generalizes previous work on hyper-gradient descent (Almeida et al., 1998; Maclaurin et al., 2015; Baydin et al., 2017) to learn a preconditioning matrix that only makes use of first-order information. Experiments show that FOP is able to improve the performance of standard deep learning optimizers on several visual classification tasks with minimal computational overhead. We also investigate the properties of the learned preconditioning matrices and perform a preliminary theoretical analysis of the algorithm. Despite this, deep neural networks and other large-scale machine learning models applied to such problems typically rely on simple variations of gradient descent to train, which is known to be highly sensitive to these difficulties.

How can we build agents that keep learning from experience, quickly and efficiently, after their initial training? Here we take inspiration from the main mechanism of learning in biological brains: synaptic plasticity, carefully tuned by evolution to produce efficient lifelong learning. We show that plasticity, just like connection weights, can be optimized by gradient descent in large (millions of parameters) recurrent networks with Hebbian plastic connections. First, recurrent plastic networks with more than two million parameters can be trained to memorize and reconstruct sets of novel, high-dimensional 1000+ pixels natural images not seen during training. Crucially, traditional non-plastic recurrent networks fail to solve this task. Furthermore, trained plastic networks can also solve generic meta-learning tasks such as the Omniglot task, with competitive results and little parameter overhead. Finally, in reinforcement learning settings, plastic networks outperform a non-plastic equivalent in a maze exploration task. We conclude that differentiable plasticity may provide a powerful novel approach to the learning-to-learn problem.

Various measures can be used to estimate bias or unfairness in a predictor. Previous work has already established that some of these measures are incompatible with each other. Here we show that, when groups differ in prevalence of the predicted event, several intuitive, reasonable measures of fairness (probability of positive prediction given occurrence or non-occurrence; probability of occurrence given prediction or non-prediction; and ratio of predictions over occurrences for each group) are all mutually exclusive: if one of them is equal among groups, the other two must differ. The only exceptions are for perfect, or trivial (always-positive or always-negative) predictors. As a consequence, any non-perfect, non-trivial predictor must necessarily be "unfair" under two out of three reasonable sets of criteria. This result readily generalizes to a wide range of well-known statistical quantities (sensitivity, specificity, false positive rate, precision, etc.), all of which can be divided into three mutually exclusive groups. Importantly, The results applies to all predictors, whether algorithmic or human. We conclude with possible ways to handle this effect when assessing and designing prediction methods.