John Chen

John Chen, Uri Wilensky

Building on Papert (1980)'s idea of children talking to computers, we propose ChatLogo, a hybrid natural-programming language interface for agent-based modeling and programming. We build upon previous efforts to scaffold ABM & P learning and recent development in leveraging large language models (LLMs) to support the learning of computational programming. ChatLogo aims to support conversations with computers in a mix of natural and programming languages, provide a more user-friendly interface for novice learners, and keep the technical system from over-reliance on any single LLM. We introduced the main elements of our design: an intelligent command center, and a conversational interface to support creative expression. We discussed the presentation format and future work. Responding to the challenges of supporting open-ended constructionist learning of ABM & P and leveraging LLMs for educational purposes, we contribute to the field by proposing the first constructionist LLM-driven interface to support computational and complex systems thinking.

John Chen, Chen Dun, Anastasios Kyrillidis

Advances in Semi-Supervised Learning (SSL) have almost entirely closed the gap between SSL and Supervised Learning at a fraction of the number of labels. However, recent performance improvements have often come \textit{at the cost of significantly increased training computation}. To address this, we propose Curriculum Batch Size (CBS), \textit{an unlabeled batch size curriculum which exploits the natural training dynamics of deep neural networks.} A small unlabeled batch size is used in the beginning of training and is gradually increased to the end of training. A fixed curriculum is used regardless of dataset, model or number of epochs, and reduced training computations is demonstrated on all settings. We apply CBS, strong labeled augmentation, Curriculum Pseudo Labeling (CPL) \citep{FlexMatch} to FixMatch \citep{FixMatch} and term the new SSL algorithm Fast FixMatch. We perform an ablation study to show that strong labeled augmentation and/or CPL do not significantly reduce training computations, but, in synergy with CBS, they achieve optimal performance. Fast FixMatch also achieves substantially higher data utilization compared to previous state-of-the-art. Fast FixMatch achieves between $2.1\times$ - $3.4\times$ reduced training computations on CIFAR-10 with all but 40, 250 and 4000 labels removed, compared to vanilla FixMatch, while attaining the same cited state-of-the-art error rate \citep{FixMatch}. Similar results are achieved for CIFAR-100, SVHN and STL-10. Finally, Fast MixMatch achieves between $2.6\times$ - $3.3\times$ reduced training computations in federated SSL tasks and online/streaming learning SSL tasks, which further demonstrate the generializbility of Fast MixMatch to different scenarios and tasks.

John Chen, Sihan Cheng, Can Gurkan et al.

Large Language Models' capacity to reason in natural language makes them uniquely promising for 4X and grand strategy games, enabling more natural human-AI gameplay interactions such as collaboration and negotiation. However, these games present unique challenges due to their complexity and long-horizon nature, while latency and cost factors may hinder LLMs' real-world deployment. Working on a classic 4X strategy game, Sid Meier's Civilization V with the Vox Populi mod, we introduce Vox Deorum, a hybrid LLM+X architecture. Our layered technical design empowers LLMs to handle macro-strategic reasoning, delegating tactical execution to subsystems (e.g., algorithmic AI or reinforcement learning AI in the future). We validate our approach through 2,327 complete games, comparing two open-source LLMs with a simple prompt against Vox Populi's enhanced AI. Results show that LLMs achieve competitive end-to-end gameplay while exhibiting play styles that diverge substantially from algorithmic AI and from each other. Our work establishes a viable architecture for integrating LLMs in commercial 4X games, opening new opportunities for game design and agentic AI research.

John Chen, Ian Berlot-Attwell, Safwan Hossain et al.

Clinical machine learning is increasingly multimodal, collected in both structured tabular formats and unstructured forms such as freetext. We propose a novel task of exploring fairness on a multimodal clinical dataset, adopting equalized odds for the downstream medical prediction tasks. To this end, we investigate a modality-agnostic fairness algorithm - equalized odds post processing - and compare it to a text-specific fairness algorithm: debiased clinical word embeddings. Despite the fact that debiased word embeddings do not explicitly address equalized odds of protected groups, we show that a text-specific approach to fairness may simultaneously achieve a good balance of performance and classical notions of fairness. We hope that our paper inspires future contributions at the critical intersection of clinical NLP and fairness. The full source code is available here: https://github.com/johntiger1/multimodal_fairness

John Chen, Sihan Cheng, Can Gurkan et al.

Evaluating strategic decision-making in LLM-based agents requires generative, competitive, and longitudinal environments, yet few benchmarks provide all three, and fewer still offer evaluation signals rich enough for long-horizon, multi-agent play. We introduce CivBench, a benchmark for LLM strategists (i.e., agentic setups) in multiplayer Civilization V. Because terminal win/loss is too sparse a signal in games spanning hundreds of turns and multiple opponents, CivBench trains models on turn-level game state to estimate victory probabilities throughout play, validated through predictive, construct, and convergent validity. Across 307 games with 7 LLMs and multiple CivBench agent conditions, we demonstrate CivBench's potential to estimate strategic capabilities as an unsaturated benchmark, reveal model-specific effects of agentic setup, and outline distinct strategic profiles not visible through outcome-only evaluation.

John Chen, Alexandros Lotsos, Sihan Cheng et al.

Qualitative analysis is critical to understanding human datasets in many social science disciplines. A central method in this process is inductive coding, where researchers identify and interpret codes directly from the datasets themselves. Yet, this exploratory approach poses challenges for meeting methodological expectations (such as ``depth'' and ``variation''), especially as researchers increasingly adopt Generative AI (GAI) for support. Ground-truth-based metrics are insufficient because they contradict the exploratory nature of inductive coding, while manual evaluation can be labor-intensive. This paper presents a theory-informed computational method for measuring inductive coding results from humans and GAI. Our method first merges individual codebooks using an LLM-enriched algorithm. It measures each coder's contribution against the merged result using four novel metrics: Coverage, Overlap, Novelty, and Divergence. Through two experiments on a human-coded online conversation dataset, we 1) reveal the merging algorithm's impact on metrics; 2) validate the metrics' stability and robustness across multiple runs and different LLMs; and 3) showcase the metrics' ability to diagnose coding issues, such as excessive or irrelevant (hallucinated) codes. Our work provides a reliable pathway for ensuring methodological rigor in human-AI qualitative analysis.

John Chen, Alexandros Lotsos, Lexie Zhao et al.

Inductive qualitative methods have been a mainstay of education research for decades, yet it takes much time and effort to conduct rigorously. Recent advances in artificial intelligence, particularly with generative AI (GAI), have led to initial success in generating inductive coding results. Like human coders, GAI tools rely on instructions to work, and how to instruct it may matter. To understand how ML/GAI approaches could contribute to qualitative coding processes, this study applied two known and two theory-informed novel approaches to an online community dataset and evaluated the resulting coding results. Our findings show significant discrepancies between ML/GAI approaches and demonstrate the advantage of our approaches, which introduce human coding processes into GAI prompts.

John Chen, Alexandros Lotsos, Grace Wang et al.

Open coding, a key inductive step in qualitative research, discovers and constructs concepts from human datasets. However, capturing extensive and nuanced aspects or "coding moments" can be challenging, especially with large discourse datasets. While some studies explore machine learning (ML)/Generative AI (GAI)'s potential for open coding, few evaluation studies exist. We compare open coding results by five recently published ML/GAI approaches and four human coders, using a dataset of online chat messages around a mobile learning software. Our systematic analysis reveals ML/GAI approaches' strengths and weaknesses, uncovering the complementary potential between humans and AI. Line-by-line AI approaches effectively identify content-based codes, while humans excel in interpreting conversational dynamics. We discussed how embedded analytical processes could shape the results of ML/GAI approaches. Instead of replacing humans in open coding, researchers should integrate AI with and according to their analytical processes, e.g., as parallel co-coders.

James T. Wilkinson, Jacob Kelter, John Chen et al.

We present a novel agent-based approach to simulating an over-the-counter (OTC) financial market in which trades are intermediated solely by market makers and agent visibility is constrained to a network topology. Dynamics, such as changes in price, result from agent-level interactions that ubiquitously occur via market maker agents acting as liquidity providers. Two additional agents are considered: trend investors use a deep convolutional neural network paired with a deep Q-learning framework to inform trading decisions by analysing price history; and value investors use a static price-target to determine their trade directions and sizes. We demonstrate that our novel inclusion of a network topology with market makers facilitates explorations into various market structures. First, we present the model and an overview of its mechanics. Second, we validate our findings via comparison to the real-world: we demonstrate a fat-tailed distribution of price changes, auto-correlated volatility, a skew negatively correlated to market maker positioning, predictable price-history patterns and more. Finally, we demonstrate that our network-based model can lend insights into the effect of market-structure on price-action. For example, we show that markets with sparsely connected intermediaries can have a critical point of fragmentation, beyond which the market forms distinct clusters and arbitrage becomes rapidly possible between the prices of different market makers. A discussion is provided on future work that would be beneficial.

John Chen, Cameron Wolfe, Anastasios Kyrillidis

Deep learning practitioners often operate on a computational and monetary budget. Thus, it is critical to design optimization algorithms that perform well under any budget. The linear learning rate schedule is considered the best budget-aware schedule, as it outperforms most other schedules in the low budget regime. On the other hand, learning rate schedules -- such as the \texttt{30-60-90} step schedule -- are known to achieve high performance when the model can be trained for many epochs. Yet, it is often not known a priori whether one's budget will be large or small; thus, the optimal choice of learning rate schedule is made on a case-by-case basis. In this paper, we frame the learning rate schedule selection problem as a combination of $i)$ selecting a profile (i.e., the continuous function that models the learning rate schedule), and $ii)$ choosing a sampling rate (i.e., how frequently the learning rate is updated/sampled from this profile). We propose a novel profile and sampling rate combination called the Reflected Exponential (REX) schedule, which we evaluate across seven different experimental settings with both SGD and Adam optimizers. REX outperforms the linear schedule in the low budget regime, while matching or exceeding the performance of several state-of-the-art learning rate schedules (linear, step, exponential, cosine, step decay on plateau, and OneCycle) in both high and low budget regimes. Furthermore, REX requires no added computation, storage, or hyperparameters.

John Chen, Qihan Wang, Anastasios Kyrillidis

The double descent curve is one of the most intriguing properties of deep neural networks. It contrasts the classical bias-variance curve with the behavior of modern neural networks, occurring where the number of samples nears the number of parameters. In this work, we explore the connection between the double descent phenomena and the number of samples in the deep neural network setting. In particular, we propose a construction which augments the existing dataset by artificially increasing the number of samples. This construction empirically mitigates the double descent curve in this setting. We reproduce existing work on deep double descent, and observe a smooth descent into the overparameterized region for our construction. This occurs both with respect to the model size, and with respect to the number epochs.

Hillary Ngai, Yoona Park, John Chen et al.

In response to the Kaggle's COVID-19 Open Research Dataset (CORD-19) challenge, we have proposed three transformer-based question-answering systems using BERT, ALBERT, and T5 models. Since the CORD-19 dataset is unlabeled, we have evaluated the question-answering models' performance on two labeled questions answers datasets \textemdash CovidQA and CovidGQA. The BERT-based QA system achieved the highest F1 score (26.32), while the ALBERT-based QA system achieved the highest Exact Match (13.04). However, numerous challenges are associated with developing high-performance question-answering systems for the ongoing COVID-19 pandemic and future pandemics. At the end of this paper, we discuss these challenges and suggest potential solutions to address them.

John Chen, Samarth Sinha, Anastasios Kyrillidis

Techniques combining multiple images as input/output have proven to be effective data augmentations for training convolutional neural networks. In this paper, we present StackMix: Each input is presented as a concatenation of two images, and the label is the mean of the two one-hot labels. On its own, StackMix rivals other widely used methods in the "Mix" line of work. More importantly, unlike previous work, significant gains across a variety of benchmarks are achieved by combining StackMix with existing Mix augmentation, effectively mixing more than two images. E.g., by combining StackMix with CutMix, test error in the supervised setting is improved across a variety of settings over CutMix, including 0.8\% on ImageNet, 3\% on Tiny ImageNet, 2\% on CIFAR-100, 0.5\% on CIFAR-10, and 1.5\% on STL-10. Similar results are achieved with Mixup.We further show that gains hold for robustness to common input corruptions and perturbations at varying severities with a 0.7\% improvement on CIFAR-100-C, by combining StackMix with AugMix over AugMix. On its own, improvements with StackMix hold across different number of labeled samples on CIFAR-100, maintaining approximately a 2\% gap in test accuracy -- down to using only 5\% of the whole dataset -- and is effective in the semi-supervised setting with a 2\% improvement with the standard benchmark $Π$-model. Finally, we perform an extensive ablation study to better understand the proposed algorithm.

Benjamin Coleman, Gaurav Gupta, John Chen et al.

Empirical risk minimization is perhaps the most influential idea in statistical learning, with applications to nearly all scientific and technical domains in the form of regression and classification models. To analyze massive streaming datasets in distributed computing environments, practitioners increasingly prefer to deploy regression models on edge rather than in the cloud. By keeping data on edge devices, we minimize the energy, communication, and data security risk associated with the model. Although it is equally advantageous to train models at the edge, a common assumption is that the model was originally trained in the cloud, since training typically requires substantial computation and memory. To this end, we propose STORM, an online sketch for empirical risk minimization. STORM compresses a data stream into a tiny array of integer counters. This sketch is sufficient to estimate a variety of surrogate losses over the original dataset. We provide rigorous theoretical analysis and show that STORM can estimate a carefully chosen surrogate loss for the least-squares objective. In an exhaustive experimental comparison for linear regression models on real-world datasets, we find that STORM allows accurate regression models to be trained.

John Chen, Vatsal Shah, Anastasios Kyrillidis

We introduce Negative Sampling in Semi-Supervised Learning (NS3L), a simple, fast, easy to tune algorithm for semi-supervised learning (SSL). NS3L is motivated by the success of negative sampling/contrastive estimation. We demonstrate that adding the NS3L loss to state-of-the-art SSL algorithms, such as the Virtual Adversarial Training (VAT), significantly improves upon vanilla VAT and its variant, VAT with Entropy Minimization. By adding the NS3L loss to MixMatch, the current state-of-the-art approach on semi-supervised tasks, we observe significant improvements over vanilla MixMatch. We conduct extensive experiments on the CIFAR10, CIFAR100, SVHN and STL10 benchmark datasets.

John Chen, Cameron Wolfe, Zhao Li et al.

Momentum is a widely used technique for gradient-based optimizers in deep learning. In this paper, we propose a decaying momentum (\textsc{Demon}) rule. We conduct the first large-scale empirical analysis of momentum decay methods for modern neural network optimization, in addition to the most popular learning rate decay schedules. Across 28 relevant combinations of models, epochs, datasets, and optimizers, \textsc{Demon} achieves the highest number of Top-1 and Top-3 finishes at 39\% and 85\% respectively, almost doubling the second-placed learning rate cosine schedule at 17\% and 60\%, respectively. \textsc{Demon} also outperforms other widely used schedulers including, but not limited to, the learning rate step schedule, linear schedule, OneCycle schedule, and exponential schedule. Compared with the widely used learning rate step schedule, \textsc{Demon} is observed to be less sensitive to parameter tuning, which is critical to training neural networks in practice. Results are demonstrated across a variety of settings and architectures, including image classification, generative models, and language models. \textsc{Demon} is easy to implement, requires no additional tuning, and incurs almost no extra computational overhead compared to the vanilla counterparts. Code is readily available.

Nicholas Ruiz, Srinivas Bangalore, John Chen

With the resurgence of chat-based dialog systems in consumer and enterprise applications, there has been much success in developing data-driven and rule-based natural language models to understand human intent. Since these models require large amounts of data and in-domain knowledge, expanding an equivalent service into new markets is disrupted by language barriers that inhibit dialog automation. This paper presents a user study to evaluate the utility of out-of-the-box machine translation technology to (1) rapidly bootstrap multilingual spoken dialog systems and (2) enable existing human analysts to understand foreign language utterances. We additionally evaluate the utility of machine translation in human assisted environments, where a portion of the traffic is processed by analysts. In English->Spanish experiments, we observe a high potential for dialog automation, as well as the potential for human analysts to process foreign language utterances with high accuracy.