arXiv subjects

Arash Ahmadi

Publications and source records attributed to Arash Ahmadi.

18 recordsLinked to original sources

arXiv2026

Adaptive Ecological Momentary Assessment with a Hybrid Language Model: Formative Expert Review and Retrospective Evaluation

Ecological momentary assessment (EMA) measures experience in daily life, but fixed questionnaires and schedules collect information of uneven value and can interrupt participants. We present and retrospectively evaluate EMA-E4B, a hybrid framework for question selection and prompt timing. Separate ridge models propose an item set and delay; a supervised Gemma 4 E4B language layer produces the final structured response and explanation. Evaluation distinguishes proxy action performance, output conformity, and formative judgments of response quality. The data contain 4,372 records from 79 participants and yield 3,516 sequential cases under a participant separated split. One involved domain expert preferred the complete hybrid response in 15 of 20 decisive comparisons, with two ties among 22 reviews. On 75 reused development cases, hybrid question utility and timing similarity were 0.832 and 0.818; the head alone reached 0.852 and 0.818. A separate 60 case comparison with untouched E4B under the same head gave action differences of -0.0031 and -0.0105. Thus, the language layer produced structured responses with action scores comparable to or slightly below the reference configurations, while the expert feedback favored the complete hybrid response. These observations establish a concrete, inspectable framework and clarify the distinct roles of action scoring and response review. Repeated adaptive administration and practical effects on measurement and participant burden remain future research.

cs.HC↗

arXiv2026

Memory-assisted squeezed light velocimetry under realistic loss and incoherent noise

We propose a velocity sensor based on a two-memory Mach--Zehnder interferometer fed by a coherent probe and squeezed vacuum, read out by balanced homodyne detection. One memory is taken as a stationary reference, while the second memory moves during storage, so that its velocity is mapped onto a differential interferometric phase at readout. The two memories are otherwise assumed identical and are described by a Gaussian write--store--read lifetime together with the associated unconditional noise floor. Using the classical Fisher information, we derive the velocity sensitivity, the transmission threshold required for a target quantum gain, and the optimum storage time. The squeezed scheme improves on equal-resource coherent homodyne within an operating window set mainly by total transmission and phase stability. For representative near-term parameters, unconditional memory noise floors up to about $10^{-1}$ photons per trial do not by themselves remove the advantage; after optimization the improvement remains at the few-percent level and is limited chiefly by loss.

quant-ph↗

arXiv2026

Enhanced LLM Reasoning by Optimizing Reward Functions with Search-Driven Reinforcement Learning

Mathematical reasoning is a key benchmark for large language models. Reinforcement learning is a standard post-training mechanism for improving the reasoning capabilities of large language models, yet performance remains sensitive to the design of the reward function that drives policy optimization. This paper introduces a search-driven framework that treats the reward specification itself as an object of optimization. The setting of interest is one in which the base model is held fixed and the reward specification is the primary remaining design lever. Candidate reward functions are generated by a frontier language model, validated automatically, screened through 500-step Group Relative Policy Optimization (GRPO) training runs on a Llama-3.2-3B-Instruct base model with Low-Rank Adaptation (LoRA), and ranked by F1 on the GSM8K test set. Ranked summaries from prior rounds are then fed back into the next round of generation. Over five rounds, the search produces 50 candidate rewards. The mean F1 rises from 0.596 in Round 1 to 0.632 in Round 5, and the top individual reward reaches F1 = 0.787. Seven ensemble configurations of top-ranked rewards are evaluated. The best ensemble achieves F1 = 0.795 (95% bootstrap CI [0.756, 0.832]) and accuracy 0.660 [0.635, 0.686], a 0.19 absolute F1 gain over a base-rewards-only GRPO baseline (F1 = 0.609). Pairwise McNemar tests with Bonferroni correction show all five-or-more-reward configurations are statistically indistinguishable at α = 0.05/21. A three-seed re-training of the best ensemble yields F1 of 0.785. A randomly drawn 5-reward control collapses to F1 = 0.047, which shows that the ranked-feedback loop, not the additive signal of having more rewards, drives the gain.

cs.CL↗

arXiv2026

Improving Heart-Focused Medical Question Answering in LLMs via Variance-Aware Rubric Rewards with GRPO

Large Language Models (LLMs) have shown strong promise in healthcare applications. Yet deploying general-purpose models in real-world settings remains difficult due to data privacy constraints, inference costs, and limited suitability for edge or on-device use. These challenges motivate the development of smaller, more efficient models that require robust post-training strategies to ensure reliable medical reasoning. In this work, we investigate Group Relative Policy Optimization (GRPO) for post-training LLMs on heart-focused medical question answering with rubric-based supervision derived from RaR-Medicine. We propose a Variance-Aware Reward Framework that extends the Explicit Aggregation and Implicit Aggregation strategies of Rubrics as Rewards by replacing weighted binary criterion aggregation and single overall Likert-style scoring with continuous analytical reward functions derived from criterion-level rubric outcomes. This formulation provides richer optimization signals for feedback that is sparse, multi-criteria, and difficult to verify automatically, and enables more stable on-policy reinforcement learning. On a held-out heart-related subset of HealthBench, our best GRPO variant improves accuracy from 0.362 to 0.502 and F1 from 0.532 to 0.668 relative to the Qwen3-14B base model, while remaining competitive with GPT-OSS-120B (0.508 accuracy, 0.674 F1). Our findings show that carefully designed rubric-based rewards provide a practical strategy for improving heart-focused medical question answering in LLMs, with potential to extend to other rubric-based tasks.

cs.CL↗

arXiv2026

MCP Bridge: A Lightweight, LLM-Agnostic RESTful Proxy for Model Context Protocol Servers

Large Language Models (LLMs) are increasingly augmented with external tools through standardized interfaces like the Model Context Protocol (MCP). However, current MCP implementations face critical limitations: they typically require local process execution through STDIO transports, making them impractical for resource-constrained environments like mobile devices, web browsers, and edge computing. We present MCP Bridge, a lightweight RESTful proxy that connects to multiple MCP servers and exposes their capabilities through a unified API. Unlike existing solutions, MCP Bridge is fully LLM-agnostic, supporting any backend regardless of vendor. The system implements a risk-based execution model with three security levels-standard execution, confirmation workflow, and Docker isolation-while maintaining backward compatibility with standard MCP clients. However, reliable execution within this framework requires models that can strictly adhere to protocol schemas. To this end, we also fine-tuned the Qwen3 4B and 8B model family on the Agent-Ark/Toucan-1.5M dataset using four Reinforcement Learning techniques: Group Relative Policy Optimization (GRPO), Dr. GRPO, Beta Normalization Policy Optimization (BNPO), and Decoupled Clip and Dynamic sAmpling Policy Optimization (DAPO). Evaluated on the MCPToolBench++ benchmark, our optimized model achieves an F1 score of 73.0% that outperforms GPT-OSS-120B (62.17%) and remains competitive with the 70B+ parameter baselines. Evaluation demonstrates that MCP Bridge successfully addresses the constraints of direct MCP connections while providing enhanced security controls and cross-platform compatibility, enabling sophisticated LLM-powered applications in previously inaccessible environments.

cs.CR↗

arXiv2025

Improving Aviation Safety Analysis: Automated HFACS Classification Using Reinforcement Learning with Group Relative Policy Optimization

Analyzing the human factors behind aviation accidents is crucial for preventing future incidents, yet traditional methods using the Human Factors Analysis and Classification System (HFACS) are limited by scalability and consistency. To address this, we introduce an automated HFACS classification framework for aviation safety analysis that utilizes Reinforcement Learning with Group Relative Policy Optimization (GRPO) to fine-tune a Llama-3.1 8B language model. Our approach incorporates a multi-component reward system tailored for aviation safety analysis and integrates synthetic data generation to overcome class imbalance in accident datasets. The resulting GRPO-optimized model achieved noticeable performance gains, including a 350% increase in exact match accuracy (from 0.0400 to 0.1800) and an improved partial match accuracy of 0.8800. Significantly, our specialized model outperforms state-of-the-art LLMs (Large Language Models), including GPT-5-mini and Gemini-2.5-fiash, on key metrics. This research also proposes exact match accuracy in multi-label HFACS classification problem as a new benchmarking methodology to evaluate the advanced reasoning capabilities of language models. Ultimately, our work validates that smaller, domain-optimized models can provide a computationally efficient and better solution for critical safety analysis. This approach makes powerful, low-latency deployment on resource-constrained edge devices feasible.

cs.CL↗

arXiv2025

Controlled Displacement of Stored Light at Room Temperature

We report the demonstration of spatially translating a stored optical pulse at room temperature over distances exceeding one optical wavelength. By implementing an interferometric scheme, we further measure the average speed of this linear translation, thus harnessing a stopped-light experiment for a sensing application. This work extends the use of quantum memories beyond quantum communication and information contexts, opening a pathway to novel methods of velocity measurements with high sensitivity.

quant-ph↗

arXiv2025

Quantum resources of quantum and classical variational methods

Variational techniques have long been at the heart of atomic, solid-state, and many-body physics. They have recently extended to quantum and classical machine learning, providing a basis for representing quantum states via neural networks. These methods generally aim to minimize the energy of a given ansätz, though open questions remain about the expressivity of quantum and classical variational ansätze. The connection between variational techniques and quantum computing, through variational quantum algorithms, offers opportunities to explore the quantum complexity of classical methods. We demonstrate how the concept of non-stabilizerness, or magic, can create a bridge between quantum information and variational techniques and we show that energy accuracy is a necessary but not always sufficient condition for accuracy in non-stabilizerness. Through systematic benchmarking of neural network quantum states, matrix product states, and variational quantum methods, we show that while classical techniques are more accurate in non-stabilizerness, not accounting for the symmetries of the system can have a severe impact on this accuracy. Our findings form a basis for a universal expressivity characterization of both quantum and classical variational methods.

quant-ph↗

arXiv2025

A Memristive Based Design of a Core Digital Circuit for Elliptic Curve Cryptography

The new emerging non-volatile memory (NVM) devices known as memristors could be the promising candidate for future digital architecture, owing to their nanoscale size and its ability to integrate with the exciting CMOS technology. In this paper, a combination of memristor devices and CMOS transistors are working together to form a hybrid CMOS-memristor circuit for XAX- Module, a core element for the finite field multiplier. The proposed design was implemented using Pt /TaOx/Ta memristor device and simulated in Cadence Virtuoso. The simulation results demonstrate the design functionality. The proposed module appears to be efficient in terms of layout area, delay and power consumption since the design utilizes the hybrid CMOS/memristor gates.

cs.CR↗

arXiv2024

Mutual information fluctuations and non-stabilizerness in random circuits

The emergence of quantum technologies has brought much attention to the characterization of quantum resources as well as the classical simulatability of quantum processes. Quantum resources, as quantified by non-stabilizerness, have in one theoretical approach been linked to a family of entropic, monotonic functions. In this work, we demonstrate both analytically and numerically a simple relationship between non-stabilizerness and information scrambling using the fluctuations of an entropy-based quantifier. Specifically, we find that the non-stabilizerness generated by a random quantum circuit is proportional to fluctuations of mutual information. Furthermore, we explore the role of non-stabilizerness in measurement-induced entanglement phase transitions. We find that the fluctuations of mutual information decrease with increasing non-stabilizerness yielding potentially easier identification of the transition point. Our work establishes a key connection between quantum resource theory, information scrambling and measurement-induced entanglement phase transitions.

quant-ph↗

arXiv2024

A Comparative Study of Sampling Methods with Cross-Validation in the FedHome Framework

This paper presents a comparative study of sampling methods within the FedHome framework, designed for personalized in-home health monitoring. FedHome leverages federated learning (FL) and generative convolutional autoencoders (GCAE) to train models on decentralized edge devices while prioritizing data privacy. A notable challenge in this domain is the class imbalance in health data, where critical events such as falls are underrepresented, adversely affecting model performance. To address this, the research evaluates six oversampling techniques using Stratified K-fold cross-validation: SMOTE, Borderline-SMOTE, Random OverSampler, SMOTE-Tomek, SVM-SMOTE, and SMOTE-ENN. These methods are tested on FedHome's public implementation over 200 training rounds with and without stratified K-fold cross-validation. The findings indicate that SMOTE-ENN achieves the most consistent test accuracy, with a standard deviation range of 0.0167-0.0176, demonstrating stable performance compared to other samplers. In contrast, SMOTE and SVM-SMOTE exhibit higher variability in performance, as reflected by their wider standard deviation ranges of 0.0157-0.0180 and 0.0155-0.0180, respectively. Similarly, the Random OverSampler method shows a significant deviation range of 0.0155-0.0176. SMOTE-Tomek, with a deviation range of 0.0160-0.0175, also shows greater stability but not as much as SMOTE-ENN. This finding highlights the potential of SMOTE-ENN to enhance the reliability and accuracy of personalized health monitoring systems within the FedHome framework.

cs.LG↗

arXiv2024

Quantifying non-stabilizerness via information scrambling

The advent of quantum technologies brought forward much attention to the theoretical characterization of the computational resources they provide. A method to quantify quantum resources is to use a class of functions called magic monotones and stabilizer entropies, which are, however, notoriously hard and impractical to evaluate for large system sizes. In recent studies, a fundamental connection between information scrambling, the magic monotone mana and 2-Renyi stabilizer entropy was established. This connection simplified magic monotone calculation, but this class of methods still suffers from exponential scaling with respect to the number of qubits. In this work, we establish a way to sample an out-of-time-order correlator that approximates magic monotones and 2-Renyi stabilizer entropy. We numerically show the relation of these sampled correlators to different non-stabilizerness measures for both qubit and qutrit systems and provide an analytical relation to 2-Renyi stabilizer entropy. Furthermore, we put forward and simulate a protocol to measure the monotonic behaviour of magic for the time evolution of local Hamiltonians.

quant-ph↗

arXiv2022

Quantum dot molecule devices with optical control of charge status and electronic control of coupling

Tunnel-coupled pairs of optically active quantum dots - quantum dot molecules (QDMs) - offer the possibility to combine excellent optical properties such as strong light-matter coupling with two-spin singlet-triplet ($S-T_0$) qubits having extended coherence times. The $S-T_0$ basis formed using two spins is inherently protected against electric and magnetic field noise. However, since a single gate voltage is typically used to stabilize the charge occupancy of the dots and control the inter-dot orbital couplings, operation of the $S-T_0$ qubits under optimal conditions remains challenging. Here, we present an electric field tunable QDM that can be optically charged with one (1h) or two holes (2h) on demand. We perform a four-phase optical and electric field control sequence that facilitates the sequential preparation of the 2h charge state and subsequently allows flexible control of the inter-dot coupling. Charges are loaded via optical pumping and electron tunnel ionization. We achieve one- and two-hole charging efficiencies of 93.5 $\pm$ 0.8 % and 80.5 $\pm$ 1.3 %, respectively. Combining efficient charge state preparation and precise setting of inter-dot coupling allows control of few-spin qubits, as would be required for on-demand generation of two-dimensional photonic cluster states or quantum transduction between microwaves and photons.

cond-mat.mes-hall↗

arXiv2020

Enhancing secure key rates of satellite QKD using a quantum dot single-photon source

Global quantum secure communication can be achieved using quantum key distribution (QKD) with orbiting satellites. Established techniques use attenuated lasers as weak coherent pulse (WCP) sources, with so-called decoy-state protocols, to generate the required single-photon-level pulses. While such approaches are elegant, they come at the expense of attainable final key due to inherent multi-photon emission, thereby constraining secure key generation over the high-loss, noisy channels expected for satellite transmissions. In this work we improve on this limitation by using true single-photon pulses generated from a semiconductor quantum dot (QD) embedded in a nanowire, possessing low multi-photon emission ($<10^{-6}$) and an extraction system efficiency of -15 dB (or 3.1%). Despite the limited efficiency, the key generated by the QD source is greater than that generated by a WCP source under identical repetition rate and link conditions representative of a satellite pass. We predict that with realistic improvements of the QD extraction efficiency to -4.0 dB (or 40%), the quantum-dot QKD protocol outperforms WCP-decoy-state QKD by almost an order of magnitude. Consequently, a QD source could allow generation of a secure key in conditions where a WCP source would simply fail, such as in the case of high channel losses. Our demonstration is the first specific use case that shows a clear benefit for QD-based single-photon sources in secure quantum communication, and has the potential to enhance the viability and efficiency of satellite-based QKD networks.

quant-ph↗

arXiv2019

Proposed Scheme to Generate Bright Entangled Photon Pairs by Application of a Quadrupole Field to a Single Quantum Dot

Entangled photon sources are crucial for quantum optics, quantum sensing and quantum communication. Semiconductor quantum dots generate on-demand entangled photon pairs via the biexciton-exciton cascade. However, the pair of photons are emitted isotropically in all directions, thus limiting the collection efficiency to a fraction of a percent. Moreover, strain and structural asymmetry in quantum dots lift the degeneracy of the intermediate exciton states in the cascade, thus degrading the measured entanglement fidelity. Here, we propose an approach for generating a pair of entangled photons from a semiconductor quantum dot by application of a quadrupole electrostatic potential. We show that the quadrupole electric field corrects for the spatial asymmetry of the excitonic wavefunction for any quantum dot dipole orientation and fully erases the fine-structure splitting without compromising the spatial overlap between electrons and holes. Our approach is compatible with nanophotonic structures such as microcavities and nanowires, thus paving the way towards a deterministic source of entangled photons with high fidelity and collection efficiency.

quant-ph↗

arXiv2016

evt_MNIST: A spike based version of traditional MNIST

Benchmarks and datasets have important role in evaluation of machine learning algorithms and neural network implementations. Traditional dataset for images such as MNIST is applied to evaluate efficiency of different training algorithms in neural networks. This demand is different in Spiking Neural Networks (SNN) as they require spiking inputs. It is widely believed, in the biological cortex the timing of spikes is irregular. Poisson distributions provide adequate descriptions of the irregularity in generating appropriate spikes. Here, we introduce a spike-based version of MNSIT (handwritten digits dataset),using Poisson distribution and show the Poissonian property of the generated streams. We introduce a new version of evt_MNIST which can be used for neural network evaluation.

cs.NE↗

arXiv2015

Optimized Implementation of Memristor-Based Full Adder by Material Implication Logic

Recently memristor-based applications and circuits are receiving an increased attention. Furthermore, memristors are also applied in logic circuit design. Material implication logic is one of the main areas with memristors. In this paper an optimized memristor-based full adder design by material implication logic is presented. This design needs 27 memristors and less area in comparison with typical CMOS-based 8-bit full adders. Also the presented full adder needs only 184 computational steps which enhance former full adder design speed by 20 percent.

cs.ET↗

arXiv2012

Biologically Inspired Spiking Neurons : Piecewise Linear Models and Digital Implementation

There has been a strong push recently to examine biological scale simulations of neuromorphic algorithms to achieve stronger inference capabilities. This paper presents a set of piecewise linear spiking neuron models, which can reproduce different behaviors, similar to the biological neuron, both for a single neuron as well as a network of neurons. The proposed models are investigated, in terms of digital implementation feasibility and costs, targeting large scale hardware implementation. Hardware synthesis and physical implementations on FPGA show that the proposed models can produce precise neural behaviors with higher performance and considerably lower implementation costs compared with the original model. Accordingly, a compact structure of the models which can be trained with supervised and unsupervised learning algorithms has been developed. Using this structure and based on a spike rate coding, a character recognition case study has been implemented and tested.

cs.LG↗