Symbolic models play a key role in cognitive science, expressing computationally precise hypotheses about how the brain implements a cognitive process. Identifying an appropriate model typically requires a great deal of effort and ingenuity on the part of a human scientist. Here, we adapt FunSearch (Romera-Paredes et al. 2024), a recently developed tool that uses Large Language Models (LLMs) in an evolutionary algorithm, to automatically discover symbolic cognitive models that accurately capture human and animal behavior. We consider datasets from three species performing a classic reward-learning task that has been the focus of substantial modeling effort, and find that the discovered programs outperform state-of-the-art cognitive models for each. The discovered programs can readily be interpreted as hypotheses about human and animal cognition, instantiating interpretable symbolic learning and decision-making algorithms. Broadly, these results demonstrate the viability of using LLM-powered program synthesis to propose novel scientific hypotheses regarding mechanisms of human and animal cognition.

Advancing planning and reasoning capabilities of Large Language Models (LLMs) is one of the key prerequisites towards unlocking their potential for performing reliably in complex and impactful domains. In this paper, we aim to demonstrate this across board games (Chess, Fischer Random / Chess960, Connect Four, and Hex), and we show that search-based planning can yield significant improvements in LLM game-playing strength. We introduce, compare and contrast two major approaches: In external search, the model guides Monte Carlo Tree Search (MCTS) rollouts and evaluations without calls to an external game engine, and in internal search, the model is trained to generate in-context a linearized tree of search and a resulting final choice. Both build on a language model pre-trained on relevant domain knowledge, reliably capturing the transition and value functions in the respective environments, with minimal hallucinations. We evaluate our LLM search implementations against game-specific state-of-the-art engines, showcasing substantial improvements in strength over the base model, and reaching Grandmaster-level performance in chess while operating closer to the human search budget. Our proposed approach, combining search with domain knowledge, is not specific to board games, hinting at more general future applications.

While Generative AI rapidly advances in various domains, generating truly creative, aesthetic, and counter-intuitive outputs remains a challenge. This paper presents an approach to tackle these difficulties in the domain of chess puzzles. We start by benchmarking Generative AI architectures, and then introduce an RL framework with novel rewards based on chess engine search statistics to overcome some of those shortcomings. The rewards are designed to enhance a puzzle's uniqueness, counter-intuitiveness, diversity, and realism. Our RL approach dramatically increases counter-intuitive puzzle generation by 10x, from 0.22\% (supervised) to 2.5\%, surpassing existing dataset rates (2.1\%) and the best Lichess-trained model (0.4\%). Our puzzles meet novelty and diversity benchmarks, retain aesthetic themes, and are rated by human experts as more creative, enjoyable, and counter-intuitive than composed book puzzles, even approaching classic compositions. Our final outcome is a curated booklet of these novel AI-generated puzzles, which is acknowledged for creativity by three world-renowned experts.

Advances in algorithmic fairness have largely omitted sexual orientation and gender identity. We explore queer concerns in privacy, censorship, language, online safety, health, and employment to study the positive and negative effects of artificial intelligence on queer communities. These issues underscore the need for new directions in fairness research that take into account a multiplicity of considerations, from privacy preservation, context sensitivity and process fairness, to an awareness of sociotechnical impact and the increasingly important role of inclusive and participatory research processes. Most current approaches for algorithmic fairness assume that the target characteristics for fairness---frequently, race and legal gender---can be observed or recorded. Sexual orientation and gender identity are prototypical instances of unobserved characteristics, which are frequently missing, unknown or fundamentally unmeasurable. This paper highlights the importance of developing new approaches for algorithmic fairness that break away from the prevailing assumption of observed characteristics.

OBJECTIVE: Multitask learning (MTL) using electronic health records allows concurrent prediction of multiple endpoints. MTL has shown promise in improving model performance and training efficiency; however, it often suffers from negative transfer - impaired learning if tasks are not appropriately selected. We introduce a sequential subnetwork routing (SeqSNR) architecture that uses soft parameter sharing to find related tasks and encourage cross-learning between them. MATERIALS AND METHODS: Using the MIMIC-III (Medical Information Mart for Intensive Care-III) dataset, we train deep neural network models to predict the onset of 6 endpoints including specific organ dysfunctions and general clinical outcomes: acute kidney injury, continuous renal replacement therapy, mechanical ventilation, vasoactive medications, mortality, and length of stay. We compare single-task (ST) models with naive multitask and SeqSNR in terms of discriminative performance and label efficiency. RESULTS: SeqSNR showed a modest yet statistically significant performance boost across 4 of 6 tasks compared with ST and naive multitasking. When the size of the training dataset was reduced for a given task (label efficiency), SeqSNR outperformed ST for all cases showing an average area under the precision-recall curve boost of 2.1%, 2.9%, and 2.1% for tasks using 1%, 5%, and 10% of labels, respectively. CONCLUSIONS: The SeqSNR architecture shows superior label efficiency compared with ST and naive multitasking, suggesting utility in scenarios in which endpoint labels are difficult to ascertain.

The hubness phenomenon has recently come into focus as an important aspect of the curse of dimensionality that affects many instance-based learning systems. It has to do with the long-tailed distribution of instance relevance within the models, where a small number of hub points dominates the analysis and influences many predictions. High data hubness is therefore often linked to poor system performance. In this paper, we re-examine several hubness-aware metric learning strategies that propose to improve system performance by reducing the expected data hubness. Instead of observing only the expected hubness degrees, our comparisons are aimed at evaluating the shape of the induced hubness degree distribution, in order to better estimate the associated hubness risk. It is revealed that the distribution of hubness degree tends to become highly skewed in many dimensions, so that many samples exhibit substantially higher hubness than expected. We argue that the long-tailed high-hubness-degree-variance metrics can be susceptible to highly detrimental high-hubness events, even if they reduce the expected hubness of the data. The experiments indicate significant differences between the compared hubness-aware metric learning approaches and show that simhubs entails the lowest overall hubness risk with increasing dimensionality. This is further shown to improve classifier stability for k-nearest neighbor classification methods.

High-dimensional data arise naturally in many domains, and have regularly presented a great challenge for traditional data mining techniques, both in terms of effectiveness and efficiency. Clustering becomes difficult due to the increasing sparsity of such data, as well as the increasing difficulty in distinguishing distances between data points. In this paper, we take a novel perspective on the problem of clustering high-dimensional data. Instead of attempting to avoid the curse of dimensionality by observing a lower dimensional feature subspace, we embrace dimensionality by taking advantage of inherently high-dimensional phenomena. More specifically, we show that hubness, i.e., the tendency of high-dimensional data to contain points (hubs) that frequently occur in k-nearest-neighbor lists of other points, can be successfully exploited in clustering. We validate our hypothesis by demonstrating that hubness is a good measure of point centrality within a high-dimensional data cluster, and by proposing several hubness-based clustering algorithms, showing that major hubs can be used effectively as cluster prototypes or as guides during the search for centroid-based cluster configurations. Experimental results demonstrate good performance of our algorithms in multiple settings, particularly in the presence of large quantities of noise. The proposed methods are tailored mostly for detecting approximately hyperspherical clusters and need to be extended to properly handle clusters of arbitrary shapes.

Most machine-learning tasks, including classification, involve dealing with high-dimensional data. It was recently shown that the phenomenon of hubness, inherent to high-dimensional data, can be exploited to improve methods based on nearest neighbors (NNs). Hubness refers to the emergence of points (hubs) that appear among the k NNs of many other points in the data, and constitute influential points for kNN classification. In this paper, we present a new probabilistic approach to kNN classification, naive hubness Bayesian k-nearest neighbor (NHBNN), which employs hubness for computing class likelihood estimates. Experiments show that NHBNN compares favorably to different variants of the kNN classifier, including probabilistic kNN (PNN) which is often used as an underlying probabilistic framework for NN classification, signifying that NHBNN is a promising alternative framework for developing probabilistic NN algorithms.