arXiv subjects

Dong-Kyum Kim

Publications and source records attributed to Dong-Kyum Kim.

10 recordsLinked to original sources

arXiv2026

Uncovering Spontaneous Physics Representations in In-Context Learning

In-context learning (ICL) lets large language models (LLMs) solve new tasks from prompts alone, across an ever-widening range of domains, yet the mechanisms underlying this ability remain poorly understood. Physical systems offer a controlled testbed for this question as they provide experimentally controllable data with structured dynamics grounded in fundamental principles. Here we study the ICL ability of LLMs, focusing on physical reasoning. Using dynamics forecasting as a proxy task, we first show that LLMs forecast physical dynamics in context, with accuracy improving as more history is provided. Analyzing the model's residual stream reveals internal activations that correlate with key physical quantities such as energy. These correlations strengthen gradually with context length, indicating that LLMs spontaneously form representations aligned with physical concepts without any physics-specific supervision. To assess whether these representations contribute to the model's predictions, we introduce a layer-wise gradient-based attribution analysis. We find that, residual directions more strongly correlated with energy also receive greater attribution to numerical predictions. This pattern is not observed for features correlated with directly observed quantities such as displacement, suggesting that the energy-related signal is not merely numerical information copied from the input. Our results broaden ICL analysis to structured physical dynamics and give a mechanistic account of how LLMs organize physical structure in context.

cs.CL↗

arXiv2026

AI Engram: In Search of Memory Traces in Artificial Intelligence

Memory formation is fundamental to intelligence, yet whether deep neural networks preserve identifiable memory traces analogous to biological memory units remains an open question. This work introduces a geometric framework to identify such "AI engrams" by formalizing the neuroscientific criteria of specificity, reactivation, sufficiency, and necessity into a constrained inverse problem. We derive a closed-form estimator that isolates individual memory traces from globally entangled parameters, and show that this biologically-derived solution corresponds to a natural gradient update on the parameter manifold. AI engrams enable surgical manipulation of learned knowledge: any subset of memories can be composed or erased through linear arithmetic, without iterative optimization. Experiments ranging from simple MLPs to LLMs demonstrate the causal validity and substantial scalability of AI engrams. Together, these results bridge theories of biological memory and artificial representation learning and offer geometric insight into how deep networks simultaneously support functional specificity within distributed storage.

cs.AI↗

arXiv2026

How Training Data Shapes the Use of Parametric and In-Context Knowledge in Language Models

Large language models leverage both parametric knowledge acquired during pretraining and in-context knowledge provided at inference time. Crucially, when these sources conflict, models arbitrate based on their internal confidence, preferring parametric knowledge for high-confidence facts while deferring to context for less familiar ones. However, the training conditions that give rise to these fundamental behaviors remain unclear. Here we conduct controlled experiments using synthetic corpora to identify the specific data properties that shape knowledge utilization. Our results reveal a counterintuitive finding: the robust, balanced use of both knowledge sources is an emergent property that requires the co-occurrence of three factors typically considered detrimental, including (i) intra-document repetition, (ii) a moderate degree of intra-document inconsistency, and (iii) a skewed knowledge distribution. We further show that these dynamics arise in real-world language model pretraining and analyze how post-training procedures reshape arbitration strategies. Together, our findings provide empirical guidance for designing training data that supports the reliable integration of parametric and in-context knowledge in language models.

cs.CL↗

arXiv2026

Erase or Hide? Suppressing Spurious Unlearning Neurons for Robust Unlearning

Large language models trained on web-scale data can memorize private or sensitive knowledge, raising significant privacy risks. Although some unlearning methods mitigate these risks, they remain vulnerable to "relearning" during subsequent training, allowing a substantial portion of forgotten knowledge to resurface. In this paper, we show that widely used unlearning methods cause shallow alignment: instead of faithfully erasing target knowledge, they generate spurious unlearning neurons that amplify negative influence to hide it. To overcome this limitation, we introduce Ssiuu, a new class of unlearning methods that employs attribution-guided regularization to prevent spurious negative influence and faithfully remove target knowledge. Experimental results confirm that our method reliably erases target knowledge and outperforms strong baselines across two practical retraining scenarios: (1) adversarial injection of private data, and (2) benign attack using an instruction-following benchmark. Our findings highlight the necessity of robust and faithful unlearning methods for safe deployment of language models.

cs.LG↗

arXiv2026

Bilinear representation mitigates reversal curse and enables consistent model editing

The reversal curse--a language model's inability to infer an unseen fact "B is A" from a learned fact "A is B"--is widely considered a fundamental limitation. We show that this is not an inherent failure but an artifact of how models encode knowledge. Our results demonstrate that training from scratch on synthetic relational knowledge graphs leads to the emergence of a bilinear relational structure within the models' hidden representations. This structure alleviates the reversal curse and facilitates inference of unseen reverse facts. Crucially, this bilinear geometry is foundational for consistent model editing: updates to a single fact propagate correctly to its reverse and logically dependent relations. In contrast, models lacking this representation suffer from the reversal curse and fail to generalize model edits, leading to logical inconsistencies. Our results establish that training on a relational knowledge dataset induces the emergence of bilinear internal representations, which in turn support language models in behaving in a logically consistent manner after editing. This suggests that the efficacy of language model editing depends not only on the choice of algorithm but on the underlying representational geometry of the knowledge itself.

cs.AI↗

arXiv2023

Multidimensional entropic bound: Estimator of entropy production for general Langevin dynamics with an arbitrary time-dependent protocol

Entropy production (EP) is a key quantity in thermodynamics, and yet measuring EP has remained a challenging task. Here we introduce an EP estimator, called multidimensional entropic bound (MEB), utilizing an ensemble of trajectories without knowing the details of a given system. MEB is a unified method in the sense that it is applicable to both overdamped and underdamped Langevin dynamics, irrespective of the time dependence of the driving protocol. In addition, MEB is computationally efficient because optimization is unnecessary. We apply our developed estimator to three physical systems driven by time-dependent protocols pertaining to experiments using optical tweezers: a dragged Brownian particle, a pulling process of a harmonic chain, and an unfolding process of an RNA hairpin. Numerical simulations confirm the validity and efficiency of our method.

cond-mat.stat-mech↗

arXiv2021

Estimating entropy production in a stochastic system with odd-parity variables

Entropy production (EP) is a central measure in nonequilibrium thermodynamics, as it can quantify the irreversibility of a process as well as its energy dissipation in special cases. Using the time-reversal asymmetry in a system's path probability distribution, many methods have been developed to estimate EP from only trajectory data. However, estimating the EP of a system with odd-parity variables, which prevails in nonequilibrium systems, has not been covered. In this study, we develop a machine learning method for estimating the EP in a stochastic system with odd-parity variables through multiple neural networks. We demonstrate our method with two systems, an underdamped bead-spring model and a one-particle odd-parity Markov jump process.

cond-mat.stat-mech↗

arXiv2021

Attaining entropy production and dissipation maps from Brownian movies via neural networks

Quantifying entropy production (EP) is essential to understand stochastic systems at mesoscopic scales, such as living organisms or biological assemblies. However, without tracking the relevant variables, it is challenging to figure out where and to what extent EP occurs from recorded time-series image data from experiments. Here, applying a convolutional neural network (CNN), a powerful tool for image processing, we develop an estimation method for EP through an unsupervised learning algorithm that calculates only from movies. Together with an attention map of the CNN's last layer, our method can not only quantify stochastic EP but also produce the spatiotemporal pattern of the EP (dissipation map). We show that our method accurately measures the EP and creates a dissipation map in two nonequilibrium systems, the bead-spring model and a network of elastic filaments. We further confirm high performance even with noisy, low spatial resolution data, and partially observed situations. Our method will provide a practical way to obtain dissipation maps and ultimately contribute to uncovering the nonequilibrium nature of complex systems.

cond-mat.stat-mech↗

arXiv2021

Deep reinforcement learning for feedback control in a collective flashing ratchet

A collective flashing ratchet transports Brownian particles using a spatially periodic, asymmetric, and time-dependent on-off switchable potential. The net current of the particles in this system can be substantially increased by feedback control based on the particle positions. Several feedback policies for maximizing the current have been proposed, but optimal policies have not been found for a moderate number of particles. Here, we use deep reinforcement learning (RL) to find optimal policies, with results showing that policies built with a suitable neural network architecture outperform the previous policies. Moreover, even in a time-delayed feedback situation where the on-off switching of the potential is delayed, we demonstrate that the policies provided by deep RL provide higher currents than the previous strategies.

cs.LG↗

arXiv2020

Learning entropy production via neural networks

This Letter presents a neural estimator for entropy production, or NEEP, that estimates entropy production (EP) from trajectories of relevant variables without detailed information on the system dynamics. For steady state, we rigorously prove that the estimator, which can be built up from different choices of deep neural networks, provides stochastic EP by optimizing the objective function proposed here. We verify the NEEP with the stochastic processes of the bead-spring and discrete flashing ratchet models, and also demonstrate that our method is applicable to high-dimensional data and can provide coarse-grained EP for Markov systems with unobservable states.

cond-mat.stat-mech↗