Large Language Models (LLMs) have transformed software engineering, but their application to physical engineering domains remains underexplored. This paper evaluates LLMs' capabilities in high-powered rocketry design through RocketBench, a benchmark connecting LLMs to high-fidelity rocket simulations. We test models on two increasingly complex design tasks: target altitude optimization and precision landing challenges. Our findings reveal that while state-of-the-art LLMs demonstrate strong baseline engineering knowledge, they struggle to iterate on their designs when given simulation results and ultimately plateau below human performance levels. However, when enhanced with reinforcement learning (RL), we show that a 7B parameter model outperforms both SoTA foundation models and human experts. This research demonstrates that RL-trained LLMs can serve as effective tools for complex engineering optimization, potentially transforming engineering domains beyond software development.

Audio-visual speech recognition (AVSR) incorporates auditory and visual modalities to improve recognition accuracy, particularly in noisy environments where audio-only speech systems are insufficient. While previous research has largely addressed audio disruptions, few studies have dealt with visual corruptions, e.g., lip occlusions or blurred videos, which are also detrimental. To address this real-world challenge, we propose CAV2vec, a novel self-supervised speech representation learning framework particularly designed to handle audio-visual joint corruption. CAV2vec employs a self-distillation approach with a corrupted prediction task, where the student model learns to predict clean targets, generated by the teacher model, with corrupted input frames. Specifically, we suggest a unimodal multi-task learning, which distills cross-modal knowledge and aligns the corrupted modalities, by predicting clean audio targets with corrupted videos, and clean video targets with corrupted audios. This strategy mitigates the dispersion in the representation space caused by corrupted modalities, leading to more reliable and robust audio-visual fusion. Our experiments on robust AVSR benchmarks demonstrate that the corrupted representation learning method significantly enhances recognition accuracy across generalized environments involving various types of corruption. Our code is available at https://github.com/sungnyun/cav2vec.

Due to the vast number of students enrolled in programming courses, there has been an increasing number of automated program repair techniques focused on introductory programming assignments (IPAs). Typically, such techniques use program clustering to take advantage of previous correct student implementations to repair a new incorrect submission. These repair techniques use clustering methods since analyzing all available correct submissions to repair a program is not feasible. However, conventional clustering methods rely on program representations based on features such as abstract syntax trees (ASTs), syntax, control flow, and data flow. This paper proposes InvAASTCluster, a novel approach for program clustering that uses dynamically generated program invariants to cluster semantically equivalent IPAs. InvAASTCluster's program representation uses a combination of the program's semantics, through its invariants, and its structure through its anonymized abstract syntax tree (AASTs). Invariants denote conditions that must remain true during program execution, while AASTs are ASTs devoid of variable and function names, retaining only their types. Our experiments show that the proposed program representation outperforms syntax-based representations when clustering a set of correct IPAs. Furthermore, we integrate InvAASTCluster into a state-of-the-art clustering-based program repair tool. Our results show that InvAASTCluster advances the current state-of-the-art when used by clustering-based repair tools by repairing around 13% more students' programs, in a shorter amount of time.

A young man with no government experience who has yet to even complete his undergraduate degree is working for Elon Musk's so-called Department of Government Efficiency (DOGE) at the Department of Housing and Urban Development (HUD) and has been tasked with using artificial intelligence to rewrite the agency's rules and regulations. Christopher Sweet was introduced to HUD employees as being originally from San Francisco and most recently a third-year at the University of Chicago, where he was studying economics and data science, in an email sent to staffers earlier this month. "I'd like to share with you that Chris Sweet has joined the HUD DOGE team with the title of special assistant, although a better title might be'Al computer programming quant analyst,'" Scott Langmack, a DOGE staffer and chief operating officer of an AI real estate company, wrote in an email widely shared within the agency and reviewed by WIRED. "With family roots from Brazil, Chris speaks Portuguese fluently. Please join me in welcoming Chris to HUD!" Sweet's primary role appears to be leading an effort to leverage artificial intelligence to review HUD's regulations, compare them to the laws on which they are based, and identify areas where rules can be relaxed or removed altogether.

Continual learning is the one of the most essential abilities for autonomous agents, which can incrementally learn daily-life skills. For this ultimate goal, a simple but powerful method, dark experience replay (DER), has been proposed recently. DER mitigates catastrophic forgetting, in which the skills acquired in the past are unintentionally forgotten, by stochastically storing the streaming data in a reservoir sampling (RS) buffer and by relearning them or retaining the past outputs for them. However, since DER considers multiple objectives, it will not function properly without appropriate weighting of them. In addition, the ability to retain past outputs inhibits learning if the past outputs are incorrect due to distribution shift or other effects. This is due to a tradeoff between memory consolidation and plasticity. The tradeoff is hidden even in the RS buffer, which gradually stops storing new data for new skills in it as data is continuously passed to it. To alleviate the tradeoff and achieve better balance, this paper proposes improvement strategies to each of DER and RS. Specifically, DER is improved with automatic adaptation of weights, block of replaying erroneous data, and correction of past outputs. RS is also improved with generalization of acceptance probability, stratification of plural buffers, and intentional omission of unnecessary data. These improvements are verified through multiple benchmarks including regression, classification, and reinforcement learning problems. As a result, the proposed methods achieve steady improvements in learning performance by balancing the memory consolidation and plasticity.

In an era increasingly shaped by decentralized knowledge ecosystems and pervasive AI technologies, fostering sustainable learner agency has become a critical educational imperative. This paper introduces a novel conceptual framework integrating Generative Artificial Intelligence (GAI) and Learning Analytics (LA) to cultivate Self - Directed Growth -- a dynamic competency enabling learner s to iteratively drive their own developmental pathways across diverse contexts. Building upon critical gaps in current Self - Directed Learning (SDL) and AI - mediated educational research, the proposed Aspire to Potentials for Learners (A2PL) model reconcept ualizes the interplay of learner aspirations, complex thinking, and summative self - assessment within GAI - supported environments. Methodological implications for future intervention designs and data analytics are discussed, positioning Self - Directed Growth as a pivotal axis for designing equitable, adaptive, and sustainable learning systems in the digital era. 1. Introduction The educational realm faces two increasingly prominent challenges that threaten to reshape the landscape of learning and development . Firstly, the traditional teacher - dominated, institution - centered environment is being eclipsed by a decentralized, ever - evolving, and technologically advanced online landscape. In this new paradigm, knowledge and skills are not poised and delivered by a single expositor, but are constantly renewed, reproduced, and reiterated through sharing and co - creation, rendering existing models of education insufficient. And the overreliance on EdTech tools, as well as information search and synthesis tools, such as Generative Artificial Intelligence (GAI), among students poses a significant challenge in the contemporary educational landscape, while there is a concerning lack of research examining whether these tools genuinely foster the development of learner agency. The integration of AI into educational practices offers a transformative opportunity to enhance learning outcomes and promote equity. According to the United Nations Educational, Scientific and Cultural Organization (UNESCO), AI has the potential to acc elerate the achievement of Sustainable Development Goal 4 (SDG 4) by improving access to quality education for all learners, regardless of their socioeconomic background (UNESCO, 2019; UNESCO, 2021). As some noted, AI facilitates access to information and online education, helping to bridge the information, skill, and educational gaps faced by disadvantaged individuals who encounter barriers to traditional learning opportunities due to time constraints, financial limitations, geographic distance, or physic al challenges (Thakkar et al., 2020; Sanabria - Z et al., 2023).

We consider generating post-hoc explanations of clusters generated from various datasets using auxiliary information which was not used by clustering algorithms. Following terminology used in previous work, we refer to the auxiliary information as tags. Our focus is on two forms of explanations, namely disjunctive form (where the explanation for a cluster consists of a set of tags) and a two-clause conjunctive normal form (CNF) explanation (where the explanation consists of two sets of tags, combined through the AND operator). We use integer linear programming (ILP) as well as heuristic methods to generate these explanations. We experiment with a variety of datasets and discuss the insights obtained from our explanations. We also present experimental results regarding the scalability of our explanation methods.

Current mainstream deep learning techniques exhibit an over-reliance on extensive training data and a lack of adaptability to the dynamic world, marking a considerable disparity from human intelligence. To bridge this gap, Few-Shot Class-Incremental Learning (FSCIL) has emerged, focusing on continuous learning of new categories with limited samples without forgetting old knowledge. Existing FSCIL studies typically use a single model to learn knowledge across all sessions, inevitably leading to the stability-plasticity dilemma. Unlike machines, humans store varied knowledge in different cerebral cortices. Inspired by this characteristic, our paper aims to develop a method that learns independent models for each session. It can inherently prevent catastrophic forgetting. During the testing stage, our method integrates Uncertainty Quantification (UQ) for model deployment. Our method provides a fresh viewpoint for FSCIL and demonstrates the state-of-the-art performance on CIFAR-100 and mini-ImageNet datasets.

Selecting high-quality pre-training data for large language models (LLMs) is crucial for enhancing their overall performance under limited computation budget, improving both training and sample efficiency. Recent advancements in file selection primarily rely on using an existing or trained proxy model to assess the similarity of samples to a target domain, such as high quality sources BookCorpus and Wikipedia. However, upon revisiting these methods, the domain-similarity selection criteria demonstrates a diversity dilemma, i.e.dimensional collapse in the feature space, improving performance on the domain-related tasks but causing severe degradation on generic performance. To prevent collapse and enhance diversity, we propose a DiverSified File selection algorithm (DiSF), which selects the most decorrelated text files in the feature space. We approach this with a classical greedy algorithm to achieve more uniform eigenvalues in the feature covariance matrix of the selected texts, analyzing its approximation to the optimal solution under a formulation of $γ$-weakly submodular optimization problem. Empirically, we establish a benchmark and conduct extensive experiments on the TinyLlama architecture with models from 120M to 1.1B parameters. Evaluating across nine tasks from the Harness framework, DiSF demonstrates a significant improvement on overall performance. Specifically, DiSF saves 98.5% of 590M training files in SlimPajama, outperforming the full-data pre-training within a 50B training budget, and achieving about 1.5x training efficiency and 5x data efficiency.

Knowledge distillation typically transfers knowledge from a teacher model to a student model by minimizing differences between their output distributions. However, existing distillation approaches largely focus on mimicking absolute probabilities and neglect the valuable relational inductive biases embedded in the teacher's relative predictions, leading to exposure bias. In this paper, we propose Group Relative Knowledge Distillation (GRKD), a novel framework that distills teacher knowledge by learning the relative ranking among classes, rather than directly fitting the absolute distribution. Specifically, we introduce a group relative loss that encourages the student model to preserve the pairwise preference orderings provided by the teacher's outputs. Extensive experiments on classification benchmarks demonstrate that GRKD achieves superior generalization compared to existing methods, especially in tasks requiring fine-grained class differentiation. Our method provides a new perspective on exploiting teacher knowledge, focusing on relational structure rather than absolute likelihood.