Mahdi Imani

Yuchen Hou, Yongshan Chen, Zhuowen Zou et al.

Federated reinforcement learning enables decentralized agents to collaboratively improve policies or value estimates without exchanging raw trajectories. However, FedAvg-style parameter averaging is not function-space consistent: when clients use heterogeneous encoders or even identical nonlinear networks, averaged parameters need not correspond to the weighted average of client value functions in any common function space. We propose FedQHD, a federated Q-learning method using hyperdimensional (random-feature) state encoders with a linear readout, so that Q-functions are nonlinear in state yet linear in trainable parameters. This linear structure enables closed-form aggregation. With a shared encoder, the function-space consensus update coincides exactly with weighted averaging of local readout matrices. With heterogeneous encoders, the server constructs a global teacher by averaging client Q-values on a shared anchor-state set, and each client compiles this teacher into its local representation via a single ridge projection. We formalize the federation gap -- the error incurred when compiling a federated teacher into a heterogeneous client representation -- relative to a client-specific oracle projection. We show that this gap decomposes into subspace misalignment, anchor-set conditioning, and regularization bias. We further identify the anchor-to-dimension ratio $m \geq D_i$ as the well-conditioned regime in which the gap reduces to a multiple of the encoder heterogeneity floor. On four continuous-state, discrete-action control benchmarks, FedQHD matches or outperforms FedAvg-style baselines and distillation-based alternatives while requiring substantially less computation, and the empirical dependence of the federation gap on encoder dimension matches our theoretical analysis.

Mahdi Imani, Roozbeh Dehghannasiri, Ulisses M. Braga-Neto et al.

Scientists are attempting to use models of ever increasing complexity, especially in medicine, where gene-based diseases such as cancer require better modeling of cell regulation. Complex models suffer from uncertainty and experiments are needed to reduce this uncertainty. Because experiments can be costly and time-consuming it is desirable to determine experiments providing the most useful information. If a sequence of experiments is to be performed, experimental design is needed to determine the order. A classical approach is to maximally reduce the overall uncertainty in the model, meaning maximal entropy reduction. A recently proposed method takes into account both model uncertainty and the translational objective, for instance, optimal structural intervention in gene regulatory networks, where the aim is to alter the regulatory logic to maximally reduce the long-run likelihood of being in a cancerous state. The mean objective cost of uncertainty (MOCU) quantifies uncertainty based on the degree to which model uncertainty affects the objective. Experimental design involves choosing the experiment that yields the greatest reduction in MOCU. This paper introduces finite-horizon dynamic programming for MOCU-based sequential experimental design and compares it to the greedy approach, which selects one experiment at a time without consideration of the full horizon of experiments. A salient aspect of the paper is that it demonstrates the advantage of MOCU-based design over the widely used entropy-based design for both greedy and dynamic-programming strategies and investigates the effect of model conditions on the comparative performances.

Yongsheng Mei, Tian Lan, Mahdi Imani et al.

Bayesian optimization (BO) is a popular global optimization scheme for sample-efficient optimization in domains with expensive function evaluations. The existing BO techniques are capable of finding a single global optimum solution. However, finding a set of global and local optimum solutions is crucial in a wide range of real-world problems, as implementing some of the optimal solutions might not be feasible due to various practical restrictions (e.g., resource limitation, physical constraints, etc.). In such domains, if multiple solutions are known, the implementation can be quickly switched to another solution, and the best possible system performance can still be obtained. This paper develops a multimodal BO framework to effectively find a set of local/global solutions for expensive-to-evaluate multimodal objective functions. We consider the standard BO setting with Gaussian process regression representing the objective function. We analytically derive the joint distribution of the objective function and its first-order derivatives. This joint distribution is used in the body of the BO acquisition functions to search for local optima during the optimization process. We introduce variants of the well-known BO acquisition functions to the multimodal setting and demonstrate the performance of the proposed framework in locating a set of local optimum solutions using multiple optimization problems.

Mohammad Alali, Mahdi Imani

A major goal in genomics is to properly capture the complex dynamical behaviors of gene regulatory networks (GRNs). This includes inferring the complex interactions between genes, which can be used for a wide range of genomics analyses, including diagnosis or prognosis of diseases and finding effective treatments for chronic diseases such as cancer. Boolean networks have emerged as a successful class of models for capturing the behavior of GRNs. In most practical settings, inference of GRNs should be achieved through limited and temporally sparse genomics data. A large number of genes in GRNs leads to a large possible topology candidate space, which often cannot be exhaustively searched due to the limitation in computational resources. This paper develops a scalable and efficient topology inference for GRNs using Bayesian optimization and kernel-based methods. Rather than an exhaustive search over possible topologies, the proposed method constructs a Gaussian Process (GP) with a topology-inspired kernel function to account for correlation in the likelihood function. Then, using the posterior distribution of the GP model, the Bayesian optimization efficiently searches for the topology with the highest likelihood value by optimally balancing between exploration and exploitation. The performance of the proposed method is demonstrated through comprehensive numerical experiments using a well-known mammalian cell-cycle network.

Yu Li, Sizhe Tang, Rongqian Chen et al.

Automated cyber defense (ACD) seeks to protect computer networks with minimal or no human intervention, reacting to intrusions by taking corrective actions such as isolating hosts, resetting services, deploying decoys, or updating access controls. However, existing approaches for ACD, such as deep reinforcement learning (RL), often face difficult exploration in complex networks with large decision/state spaces and thus require an expensive amount of samples. Inspired by the need to learn sample-efficient defense policies, we frame ACD in CAGE Challenge 4 (CAGE-4 / CC4) as a context-based partially observable Markov decision problem and propose a planning-centric defense policy based on Monte Carlo Tree Search (MCTS). It explicitly models the exploration-exploitation tradeoff in ACD and uses statistical sampling to guide exploration and decision making. We make novel use of graph neural networks (GNNs) to embed observations from the network as attributed graphs, to enable permutation-invariant reasoning over hosts and their relationships. To make our solution practical in complex search spaces, we guide MCTS with learned graph embeddings and priors over graph-edit actions, combining model-free generalization and policy distillation with look-ahead planning. We evaluate the resulting agent on CC4 scenarios involving diverse network structures and adversary behaviors, and show that our search-guided, graph-embedding-based planning improves defense reward and robustness relative to state-of-the-art RL baselines.

Sungheon Jeong, Ryozo Masukawa, Jihong Park et al.

While recent Large Vision-Language Models (LVLMs) exhibit strong multimodal reasoning abilities, they often produce ungrounded or hallucinated responses because they rely too heavily on linguistic priors instead of visual evidence. This limitation highlights the absence of a quantitative measure of how much these models actually use visual information during reasoning. We propose Draft and Refine (DnR), an agent framework driven by a question-conditioned utilization metric. The metric quantifies the model's reliance on visual evidence by first constructing a query-conditioned relevance map to localize question-specific cues and then measuring dependence through relevance-guided probabilistic masking. Guided by this metric, the DnR agent refines its initial draft using targeted feedback from external visual experts. Each expert's output (such as boxes or masks) is rendered as visual cues on the image, and the model is re-queried to select the response that yields the largest improvement in utilization. This process strengthens visual grounding without retraining or architectural changes. Experiments across VQA and captioning benchmarks show consistent accuracy gains and reduced hallucination, demonstrating that measuring visual utilization provides a principled path toward more interpretable and evidence-driven multimodal agent systems.

Shu Hong, Yongsheng Mei, Mahdi Imani et al.

Bayesian optimization (BO) is a powerful framework for optimizing expensive black-box objectives, yet extending it to graph-structured domains remains challenging due to the discrete and combinatorial nature of graphs. Existing approaches often rely on either full graph topology-impractical for large or partially observed graphs-or incremental exploration, which can lead to slow convergence. We introduce a scalable framework for global optimization over graphs that employs low-rank spectral representations to build Gaussian process (GP) surrogates from sparse structural observations. The method jointly infers graph structure and node representations through learnable embeddings, enabling efficient global search and principled uncertainty estimation even with limited data. We also provide theoretical analysis establishing conditions for accurate recovery of underlying graph structure under different sampling regimes. Experiments on synthetic and real-world datasets demonstrate that our approach achieves faster convergence and improved optimization performance compared to prior methods.

Zuyuan Zhang, Mahdi Imani, Tian Lan

Real-world reinforcement learning is often \emph{nonstationary}: rewards and dynamics drift, accelerate, oscillate, and trigger abrupt switches in the optimal action. Existing theory often represents nonstationarity with coarse-scale models that measure \emph{how much} the environment changes, not \emph{how} it changes locally -- even though acceleration and near-ties drive tracking error and policy chattering. We take a geometric view of nonstationary discounted Markov Decision Processes (MDPs) by modeling the environment as a differentiable homotopy path and tracking the induced motion of the optimal Bellman fixed point. This yields a length-curvature-kink signature of intrinsic complexity: cumulative drift, acceleration/oscillation, and action-gap-induced nonsmoothness. We prove a solver-agnostic path-integral stability bound and derive gap-safe feasible regions that certify local stability away from switch regimes. Building on these results, we introduce \textit{Homotopy-Tracking RL (HT-RL)} and \textit{HT-MCTS}, lightweight wrappers that estimate replay-based proxies of length, curvature, and near-tie proximity online and adapt learning or planning intensity accordingly. Experiments show improved tracking and dynamic regret over matched static baselines, with the largest gains in oscillatory and switch-prone regimes.

Sungheon Jeong, Ryozo Masukawa, Sanggeon Yun et al.

Language environments such as web browsers, code terminals, and interactive simulations emit raw text rather than states, and provide none of the runtime structure that MDP analysis requires. No explicit state space, no observation-to-state mapping, no certified transitions, and no termination criterion. We introduce the State-Centric Decision Process (SDP), a runtime framework that constructs these missing inputs by having the agent build them, predicate by predicate, as it acts. At each step the agent commits to a natural-language predicate describing how the world should look, takes an action to make it true, and checks the observation against it. Predicates that pass become certified states, and the resulting trajectory carries the four objects language environments do not provide, namely a task-induced state space, an observation-to-state mapping, certified transitions, and a termination criterion. We evaluate SDP on five benchmarks spanning planning, scientific exploration, web reasoning, and multi-hop question answering. SDP achieves the best training-free results on all five, with the advantage widening as the horizon grows. The certified trajectories additionally support analyses unavailable to reactive agents, including per-predicate credit assignment, failure localization, partial-progress measurement, and modular operator replacement.

Zuyuan Zhang, Sizhe Tang, Mahdi Imani et al.

General-sum multi-agent learning is often governed by a stacked update field in which each agent's policy update changes the optimization landscape faced by the others. This coupling can entangle an integrable component of collective improvement with cyclic interaction dynamics, leading to slow or unstable multi-agent learning. Existing approaches, such as regularization, credit assignment, and consensus methods, stabilize MARL through local or algorithmic modifications; HPML complements them by projecting the joint update field onto a metric-gradient component. We introduce \textbf{HPML} (\textbf{H}odge-\textbf{P}rojected \textbf{M}ulti-agent \textbf{L}earning), which views the joint update field of a multi-agent system as an element of an $L^2$ space of vector fields and computes a Hodge-type projection onto the closest metric-gradient potential flow. HPML follows the projected component as the update direction, yielding the closest metric-gradient field under the chosen metric and sampling measure. The projection is defined variationally, characterized by a Poisson-type equation, and implemented through graph-based and amortized neural realizations that recover projected directions from samples. We show that the projected dynamics admit a Lyapunov potential and yield equilibrium-gap bounds with an explicit additive non-potentiality term. Controlled experiments validate the geometric mechanism, and CTDE benchmarks show improved stability and normalized return when HPML is used as a plug-in projection layer in MARL pipelines.

Fei Xu Yu, Zuyuan Zhang, Mahdi Imani et al.

In structured decision-making workflows such as form filling, compliance checking, and maintenance reporting, LLM outputs must be locally correct, globally consistent, and auditable against task-specific rules. Existing refinement methods often rely on heuristic debate, self-play, or LLM-generated supervision, creating a second-order assurance problem. We propose DPA-GRPO (Dual Paired-Action Group-Relative Policy Optimization), a paired-action training method for a two-player generator--verifier game with structured verifier interventions. The generator proposes outputs and may revise them when challenged; the verifier either remains silent or raises a safety assurance case (SAC) containing a claim, argument, and evidence. These SAC/no-SAC and KEEP/REVISE decisions induce paired counterfactual action groups, which DPA-GRPO uses for role-specific KL-regularized GRPO updates. We analyze the unregularized game and show that positive probability on strictly lower-reward intervention or revision actions creates a profitable unilateral deviation. Under standard stochastic-approximation assumptions, DPA-GRPO tracks the corresponding game ODE, whose isolated asymptotically stable limit points are stationary and candidate local equilibria under role-wise local optimality. Experiments on TaxCalcBench TY24 show that DPA-GRPO improves structured decision accuracy over zero-shot generation and generator-only RL baselines across Qwen3-4B and Qwen3-8B. Training increases correct silent acceptance, reduces missed errors, and improves calibrated revision behavior, indicating gains for both generator and verifier.

Sizhe Tang, Zuyuan Zhang, Mahdi Imani et al.

Monte Carlo Tree Search (MCTS) scales poorly in cooperative multi-agent domains because expansion must consider an exponentially large set of joint actions, severely limiting exploration under realistic search budgets. We propose NonZero, which keeps multi-agent MCTS tractable by running surrogate-guided selection over a low-dimensional nonlinear representation using an interaction-guided proposal rule, instead of directly exploring the full joint-action space. Our exploration uses an interaction score: single-agent deviations are ranked by predicted gain, while two-agent deviations are scored by a mixed-difference measure that reveals coordination benefits even when no single agent can improve alone. We formalize candidate proposal as a bandit problem over local deviations and derive a proposal rule, NonZero, with a sublinear local-regret guarantee for reaching approximate graph-local optima without enumerating the joint-action space. Empirically, NonZero improves sample efficiency and final performance on MatGame, SMAC, and SMACv2 relative to strong model-based and model-free baselines under matched search budgets.

Zuyuan Zhang, Hanhan Zhou, Mahdi Imani et al.

With the advancements of artificial intelligence (AI), we're seeing more scenarios that require AI to work closely with other agents, whose goals and strategies might not be known beforehand. However, existing approaches for training collaborative agents often require defined and known reward signals and cannot address the problem of teaming with unknown agents that often have latent objectives/rewards. In response to this challenge, we propose teaming with unknown agents framework, which leverages kernel density Bayesian inverse learning method for active goal deduction and utilizes pre-trained, goal-conditioned policies to enable zero-shot policy adaptation. We prove that unbiased reward estimates in our framework are sufficient for optimal teaming with unknown agents. We further evaluate the framework of redesigned multi-agent particle and StarCraft II micromanagement environments with diverse unknown agents of different behaviors/rewards. Empirical results demonstrate that our framework significantly advances the teaming performance of AI and unknown agents in a wide range of collaborative scenarios.

Zeyu Fang, Tian Lan, Mahdi Imani

Joint planning through language-based interactions is a key area of human-AI teaming. Planning problems in the open world often involve various aspects of incomplete information and unknowns, e.g., objects involved, human goals/intents -- thus leading to knowledge gaps in joint planning. We consider the problem of discovering optimal interaction strategies for AI agents to actively elicit human inputs in object-driven planning. To this end, we propose Minimal Information Neuro-Symbolic Tree (MINT) to reason about the impact of knowledge gaps and leverage self-play with MINT to optimize the AI agent's elicitation strategies and queries. More precisely, MINT builds a symbolic tree by making propositions of possible human-AI interactions and by consulting a neural planning policy to estimate the uncertainty in planning outcomes caused by remaining knowledge gaps. Finally, we leverage LLM to search and summarize MINT's reasoning process and curate a set of queries to optimally elicit human inputs for best planning performance. By considering a family of extended Markov decision processes with knowledge gaps, we analyze the return guarantee for a given MINT with active human elicitation. Our evaluation on three benchmarks involving unseen/unknown objects of increasing realism shows that MINT-based planning attains near-expert returns by issuing a limited number of questions per task while achieving significantly improved rewards and success rates.

Rongqian Chen, Allison Andreyev, Yanming Xiu et al.

Augmented Reality (AR) enriches perception by overlaying virtual elements on the physical world. Due to its growing popularity, cognitive attacks that alter AR content to manipulate users' semantic perception have received increasing attention. Existing detection methods often focus on visual changes, which are restricted to pixel- or image-level processing and lack semantic reasoning capabilities, or they rely on pre-trained vision-language models (VLMs), which function as black-box approaches with limited interpretability. In this paper, we present CADAR, a novel neurosymbolic approach for cognitive attack detection in AR. It fuses multimodal vision-language inputs using neural VLMs to obtain a symbolic perception-graph representation, incorporating prior knowledge, salience weighting, and temporal correlations. The model then enables particle-filter based statistical reasoning -- a sequential Monte Carlo method -- to detect cognitive attacks. Thus, CADAR inherits the adaptability of pre-trained VLM and the interpretability and reasoning rigor of particle filtering. Experiments on an extended AR cognitive attack dataset show accuracy improvements of up to 10.7% over strong baselines on challenging AR attack scenarios, underscoring the promise of neurosymbolic methods for effective and interpretable cognitive attack detection.

Zeyu Fang, Yuxin Lin, Cheng Liu et al.

Effective human-robot collaboration in open-world environments requires joint planning under uncertain conditions. However, existing approaches often treat humans as passive supervisors, preventing autonomous agents from becoming human-like teammates that can actively model teammate behaviors, reason about knowledge gaps, query, and elicit responses through communication to resolve uncertainties. To address these limitations, we propose a unified human-robot joint planning system designed to tackle dual sources of uncertainty: task-relevant knowledge gaps and latent human intent. Our system operates in two complementary modes. First, an uncertainty-mitigation joint planning module enables two-way conversations to resolve semantic ambiguity and object uncertainty. It utilizes an LLM-assisted active elicitation mechanism and a hypothesis-augmented A^* search, subsequently computing an optimal querying policy via dynamic programming to minimize interaction and verification costs. Second, a real-time intent-aware collaboration module maintains a probabilistic belief over the human's latent task intent via spatial and directional cues, enabling dynamic, coordination-aware task selection for agents without explicit communication. We validate the proposed system in both Gazebo simulations and real-world UAV deployments integrated with a Vision-Language Model (VLM)-based 3D semantic perception pipeline. Experimental results demonstrate that the system significantly cuts the interaction cost by 51.9% in uncertainty-mitigation planning and reduces the task execution time by 25.4% in intent-aware cooperation compared to the baselines.

Rongqian Chen, Shu Hong, Rifatul Islam et al.

Augmented reality (AR) systems are increasingly deployed in tactical environments, but their reliance on seamless human-computer interaction makes them vulnerable to cognitive attacks that manipulate a user's perception and severely compromise user decision-making. To address this challenge, we introduce the Perception Graph, a novel model designed to reason about human perception within these systems. Our model operates by first mimicking the human process of interpreting key information from an MR environment and then representing the outcomes using a semantically meaningful structure. We demonstrate how the model can compute a quantitative score that reflects the level of perception distortion, providing a robust and measurable method for detecting and analyzing the effects of such cognitive attacks.

Zeyu Fang, Zuyuan Zhang, Mahdi Imani et al.

Model-based offline reinforcement learning is brittle under distribution shift: policy improvement drives rollouts into state--action regions weakly supported by the dataset, where compounding model error yields severe value overestimation. We propose Manifold-Constrained Energy-based Transition Models (MC-ETM), which train conditional energy-based transition models using a manifold projection--diffusion negative sampler. MC-ETM learns a latent manifold of next states and generates near-manifold hard negatives by perturbing latent codes and running Langevin dynamics in latent space with the learned conditional energy, sharpening the energy landscape around the dataset support and improving sensitivity to subtle out-of-distribution deviations. For policy optimization, the learned energy provides a single reliability signal: rollouts are truncated when the minimum energy over sampled next states exceeds a threshold, and Bellman backups are stabilized via pessimistic penalties based on Q-value-level dispersion across energy-guided samples. We formalize MC-ETM through a hybrid pessimistic MDP formulation and derive a conservative performance bound separating in-support evaluation error from truncation risk. Empirically, MC-ETM improves multi-step dynamics fidelity and yields higher normalized returns on standard offline control benchmarks, particularly under irregular dynamics and sparse data coverage.

Zeyu Fang, Beomyeol Yu, Cheng Liu et al.

Human-AI joint planning in Unmanned Aerial Vehicles (UAVs) typically relies on control handover when facing environmental uncertainties, which is often inefficient and cognitively demanding for non-expert operators. To address this, we propose a novel framework that shifts the collaboration paradigm from control takeover to active information elicitation. We introduce the Minimal Information Neuro-Symbolic Tree (MINT), a reasoning mechanism that explicitly structures knowledge gaps regarding obstacles and goals into a queryable format. By leveraging large language models, our system formulates optimal binary queries to resolve specific ambiguities with minimal human interaction. We demonstrate the efficacy of this approach through a comprehensive workflow integrating a vision-language model for perception, voice interfaces, and a low-level UAV control module in both high-fidelity NVIDIA Isaac simulations and real-world deployments. Experimental results show that our method achieves a significant improvement in the success rate for complex search-and-rescue tasks while significantly reducing the frequency of human interaction compared to exhaustive querying baselines.

Ryozo Masukawa, Sanggeon Yun, Hyunwoo Oh et al.

Recent progress in reinforcement learning with verifiable rewards (RLVR) shows that small, specialized language models (SLMs) can exhibit structured reasoning without relying on large monolithic LLMs. We introduce soft hidden-state collaboration, where multiple heterogeneous frozen SLM experts are integrated through their internal representations via a trainable attention interface. Experiments on Reasoning Gym and GSM8K show that this latent integration is competitive with strong single-model RLVR baselines. Ablations further reveal a dual mechanism of expert utilization: for simpler arithmetic domains, performance gains can largely be explained by static expert preferences, whereas more challenging settings induce increasingly concentrated and structured expert attention over training, indicating emergent specialization in how the router connects to relevant experts. Overall, hidden-state collaboration provides a compact mechanism for leveraging frozen experts, while offering an observational window into expert utilization patterns and their evolution under RLVR.

Shuilian Xie, Mahdi Imani, Edward R. Dougherty et al.

Classical discriminant analysis assumes identically distributed training data, yet in many applications observations are collected over time and the class-conditional distributions drift. This population drift renders stationary classifiers unreliable. We propose a principled, model-based framework that embeds discriminant analysis within state-space models to obtain nonstationary linear discriminant analysis (NSLDA) and nonstationary quadratic discriminant analysis (NSQDA). For linear-Gaussian dynamics, we adapt Kalman smoothing to handle multiple samples per time step and develop two practical extensions: (i) an expectation-maximization (EM) approach that jointly estimates unknown system parameters, and (ii) a Gaussian mixture model (GMM)-Kalman method that simultaneously recovers unobserved time labels and parameters, a scenario common in practice. To address nonlinear or non-Gaussian drift, we employ particle smoothing to estimate time-varying class centroids, yielding fully nonstationary discriminant rules. Extensive simulations demonstrate consistent improvements over stationary linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), and support vector machine (SVM) baselines, with robustness to noise, missing data, and class imbalance. This paper establishes a unified and data-efficient foundation for discriminant analysis under temporal distribution shift.

Yongsheng Mei, Mahdi Imani, Tian Lan

Bayesian optimization (BO) has established itself as a leading strategy for efficiently optimizing expensive-to-evaluate functions. Existing BO methods mostly rely on Gaussian process (GP) surrogate models and are not applicable to (doubly-stochastic) Gaussian Cox processes, where the observation process is modulated by a latent intensity function modeled as a GP. In this paper, we propose a novel maximum a posteriori inference of Gaussian Cox processes. It leverages the Laplace approximation and change of kernel technique to transform the problem into a new reproducing kernel Hilbert space, where it becomes more tractable computationally. It enables us to obtain both a functional posterior of the latent intensity function and the covariance of the posterior, thus extending existing works that often focus on specific link functions or estimating the posterior mean. Using the result, we propose a BO framework based on the Gaussian Cox process model and further develop a Nyström approximation for efficient computation. Extensive evaluations on various synthetic and real-world datasets demonstrate significant improvement over state-of-the-art inference solutions for Gaussian Cox processes, as well as effective BO with a wide range of acquisition functions designed through the underlying Gaussian Cox process model.

Mahdi Imani, Ulisses Braga-Neto

This paper is concerned with the problem of stochastic control of gene regulatory networks (GRNs) observed indirectly through noisy measurements and with uncertainty in the intervention inputs. The partial observability of the gene states and uncertainty in the intervention process are accounted for by modeling GRNs using the partially-observed Boolean dynamical system (POBDS) signal model with noisy gene expression measurements. Obtaining the optimal infinite-horizon control strategy for this problem is not attainable in general, and we apply reinforcement learning and Gaussian process techniques to find a near-optimal solution. The POBDS is first transformed to a directly-observed Markov Decision Process in a continuous belief space, and the Gaussian process is used for modeling the cost function over the belief and intervention spaces. Reinforcement learning then is used to learn the cost function from the available gene expression data. In addition, we employ sparsification, which enables the control of large partially-observed GRNs. The performance of the resulting algorithm is studied through a comprehensive set of numerical experiments using synthetic gene expression data generated from a melanoma gene regulatory network.