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.

We study the problem of clustering networks whose nodes have imputed or physical positions in a single dimension, for example prestige hierarchies or the similarity dimension of hyperbolic embeddings. Existing algorithms, such as the critical gap method and other greedy strategies, only offer approximate solutions to this problem. Here, we introduce a dynamic programming approach that returns provably optimal solutions in polynomial time -- O(n^2) steps -- for a broad class of clustering objectives. We demonstrate the algorithm through applications to synthetic and empirical networks and show that it outperforms existing heuristics by a significant margin, with a similar execution time.

Differentiable Search Indices (DSIs) encode a corpus of documents in model parameters and use the same model to answer user queries directly. Despite the strong performance of DSI models, deploying them in situations where the corpus changes over time is computationally expensive because reindexing the corpus requires re-training the model. In this work, we introduce DSI++, a continual learning challenge for DSI to incrementally index new documents while being able to answer queries related to both previously and newly indexed documents. Across different model scales and document identifier representations, we show that continual indexing of new documents leads to considerable forgetting of previously indexed documents. We also hypothesize and verify that the model experiences forgetting events during training, leading to unstable learning. To mitigate these issues, we investigate two approaches. The first focuses on modifying the training dynamics. Flatter minima implicitly alleviate forgetting, so we optimize for flatter loss basins and show that the model stably memorizes more documents ($+12\%$). Next, we introduce a generative memory to sample pseudo-queries for documents and supplement them during continual indexing to prevent forgetting for the retrieval task. Extensive experiments on novel continual indexing benchmarks based on Natural Questions (NQ) and MS MARCO demonstrate that our proposed solution mitigates forgetting significantly. Concretely, it improves the average Hits@10 by $+21.1\%$ over competitive baselines for NQ and requires $6$ times fewer model updates compared to re-training the DSI model for incrementally indexing five corpora in a sequence.

Continual zero-shot learning involves learning seen classes incrementally while improving the ability to recognize unseen or yet-to-be-seen classes. It has a broad range of potential applications in real-world vision tasks, such as accelerating species discovery. However, in these scenarios, the changes in environmental conditions cause shifts in the presentation of captured images, which we refer to as domain shift, and adds complexity to the tasks. In this paper, we introduce Domain Aware Continual Zero-Shot Learning (DACZSL), a task that involves visually recognizing images of unseen categories in unseen domains continually. To address the challenges of DACZSL, we propose a Domain-Invariant Network (DIN). We empoly a dual network structure to learn factorized features to alleviate forgetting, where consists of a global shared net for domian-invirant and task-invariant features, and per-task private nets for task-specific features. Furthermore, we introduce a class-wise learnable prompt to obtain better class-level text representation, which enables zero-shot prediction of future unseen classes. To evaluate DACZSL, we introduce two benchmarks: DomainNet-CZSL and iWildCam-CZSL. Our results show that DIN significantly outperforms existing baselines and achieves a new state-of-the-art.

Optimizing proper loss functions is popularly believed to yield predictors with good calibration properties; the intuition being that for such losses, the global optimum is to predict the ground-truth probabilities, which is indeed calibrated. However, typical machine learning models are trained to approximately minimize loss over restricted families of predictors, that are unlikely to contain the ground truth. Under what circumstances does optimizing proper loss over a restricted family yield calibrated models? What precise calibration guarantees does it give? In this work, we provide a rigorous answer to these questions. We replace the global optimality with a local optimality condition stipulating that the (proper) loss of the predictor cannot be reduced much by post-processing its predictions with a certain family of Lipschitz functions. We show that any predictor with this local optimality satisfies smooth calibration as defined in Kakade-Foster (2008), B{\l}asiok et al. (2023). Local optimality is plausibly satisfied by well-trained DNNs, which suggests an explanation for why they are calibrated from proper loss minimization alone. Finally, we show that the connection between local optimality and calibration error goes both ways: nearly calibrated predictors are also nearly locally optimal.

The notion of causality assumes a paramount position within the realm of human cognition. Over the past few decades, there has been significant advancement in the domain of causal effect estimation across various disciplines, including but not limited to computer science, medicine, economics, and industrial applications. Given the continued advancements in deep learning methodologies, there has been a notable surge in its utilization for the estimation of causal effects using counterfactual data. Typically, deep causal models map the characteristics of covariates to a representation space and then design various objective functions to estimate counterfactual data unbiasedly. Different from the existing surveys on causal models in machine learning, this review mainly focuses on the overview of the deep causal models, and its core contributions are as follows: 1) we cast insight on a comprehensive overview of deep causal models from both timeline of development and method classification perspectives; 2) we outline some typical applications of causal effect estimation to industry; 3) we also endeavor to present a detailed categorization and analysis on relevant datasets, source codes and experiments.

In this paper, we propose a generic algorithm to train machine learning-based subgrid parametrizations online, i.e., with a posteriori loss functions, but for non-differentiable numerical solvers. The proposed approach leverages a neural emulator to approximate the reduced state-space solver, which is then used to allow gradient propagation through temporal integration steps. We apply this methodology on a single layer quasi-geostrophic system with topography, known to be highly unstable in around 500 temporal iterations with offline strategies. Using our algorithm, we are able to train a parametrization that recovers most of the benefits of online strategies without having to compute the gradient of the original solver. It is demonstrated that training the neural emulator and parametrization components separately with different loss quantities is necessary in order to minimize the propagation of approximation biases. Experiments on emulator architectures with different complexities also indicates that emulator performance is key in order to learn an accurate parametrization. This work is a step towards learning parametrization with online strategies for realistic climate models.

Deep reinforcement learning (RL) provides powerful methods for training optimal sequential decision-making agents. As collecting real-world interactions can entail additional costs and safety risks, the common paradigm of sim2real conducts training in a simulator, followed by real-world deployment. Unfortunately, RL agents easily overfit to the choice of simulated training environments, and worse still, learning ends when the agent masters the specific set of simulated environments. In contrast, the real world is highly open-ended, featuring endlessly evolving environments and challenges, making such RL approaches unsuitable. Simply randomizing over simulated environments is insufficient, as it requires making arbitrary distributional assumptions and can be combinatorially less likely to sample specific environment instances that are useful for learning. An ideal learning process should automatically adapt the training environment to maximize the learning potential of the agent over an open-ended task space that matches or surpasses the complexity of the real world. This thesis develops a class of methods called Unsupervised Environment Design (UED), which aim to produce such open-ended processes. Given an environment design space, UED automatically generates an infinite sequence or curriculum of training environments at the frontier of the learning agent's capabilities. Through extensive empirical studies and theoretical arguments founded on minimax-regret decision theory and game theory, the findings in this thesis show that UED autocurricula can produce RL agents exhibiting significantly improved robustness and generalization to previously unseen environment instances. Such autocurricula are promising paths toward open-ended learning systems that achieve more general intelligence by continually generating and mastering additional challenges of their own design.

The prevalence of online content has led to the widespread adoption of recommendation systems (RSs), which serve diverse purposes such as news, advertisements, and e-commerce recommendations. Despite their significance, data scarcity issues have significantly impaired the effectiveness of existing RS models and hindered their progress. To address this challenge, the concept of knowledge transfer, particularly from external sources like pre-trained language models, emerges as a potential solution to alleviate data scarcity and enhance RS development. However, the practice of knowledge transfer in RSs is intricate. Transferring knowledge between domains introduces data disparities, and the application of knowledge transfer in complex RS scenarios can yield negative consequences if not carefully designed. Therefore, this article contributes to this discourse by addressing the implications of data scarcity on RSs and introducing various strategies, such as data augmentation, self-supervised learning, transfer learning, broad learning, and knowledge graph utilization, to mitigate this challenge. Furthermore, it delves into the challenges and future direction within the RS domain, offering insights that are poised to facilitate the development and implementation of robust RSs, particularly when confronted with data scarcity. We aim to provide valuable guidance and inspiration for researchers and practitioners, ultimately driving advancements in the field of RS.

Learning from unstructured and uncurated data has become the dominant paradigm for generative approaches in language and vision. Such unstructured and unguided behavior data, commonly known as play, is also easier to collect in robotics but much more difficult to learn from due to its inherently multimodal, noisy, and suboptimal nature. In this paper, we study this problem of learning goal-directed skill policies from unstructured play data which is labeled with language in hindsight. Specifically, we leverage advances in diffusion models to learn a multi-task diffusion model to extract robotic skills from play data. Using a conditional denoising diffusion process in the space of states and actions, we can gracefully handle the complexity and multimodality of play data and generate diverse and interesting robot behaviors. To make diffusion models more useful for skill learning, we encourage robotic agents to acquire a vocabulary of skills by introducing discrete bottlenecks into the conditional behavior generation process. In our experiments, we demonstrate the effectiveness of our approach across a wide variety of environments in both simulation and the real world. Results visualizations and videos at https://play-fusion.github.io