MAMajid Nili Ahmadabadi

Majid Nili Ahmadabadi

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Majid Nili Ahmadabadi has a total of 102 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Fourier-Gated Transformer and MLP-Enhanced REV: Noise-Robust Phased-Array Antenna Calibration Using Power-Only Measurements

    by Zahra Sarayloo, Nasser Masoumi, Mohammad Reza Nezhadahmadi, Majid Nili Ahmadabadi

    Accurate calibration of active phased array antennas (APAAs) is essential during field operation, but phase-based methods are often costly and challenging due to the need for precise phase measurement equipment. To enable efficient online calibration using only output power measurements, we build on the rotating element electric field vector (REV) method and propose three neural network-based approaches. Each approach relies on N power samples obtained by sweeping the phase shifter (e.g., 32 samples for a 5-bit shifter). The first approach, multilayer perceptron neural-network enhanced REV (MLPeREV), uses an MLP to estimate the REV input parameters ϕ and r from the power sequence and then applies the REV equations to compute phase errors. The second approach, transformer neural network-based phase error estimation (Tran-PhEr), replaces explicit parameter estimation with an encoder-only transformer that directly processes the raw power sequence and predicts phase errors via self-attention. The third approach, Fourier-gated transformer neural network-based phase error estimation (FGTran-PhEr), augments Tran-PhEr with a Fourier transform layer and Gaussian gating to focus attention on the sequence’s maxima and minima. This yields noise-robust predictions and reduces the need for repeated measurements. These transformer-based approaches use only the REV measurement procedure, not the REV equation directly. We evaluate the three approaches in simulation and experimentation on a 1024-antenna planar phased array operating at 18–21 GHz. Experimental validation was conducted on a representative C-module (16 antennas) of the 1024-element array, with results generalizable to the full aperture. MLPeREV achieves a mean absolute error (MAE) and root-mean-square error (RMSE) of 2.30° and 2.80°, respectively, in simulation, and 4.15° and 5.07° in experimentation. Tran-PhEr yields an MAE of 1.91° (simulation) and 2.86° (experimentation), and an RMSE of 2.46° (simulation) and 3.67° (experimentation), converging in 37 epochs. FGTran-PhEr outperforms both, with an MAE of 1.36° (simulation) and 1.75° (experimentation), and an RMSE of 1.93° (simulation) and 2.48° (experimentation), converging in 24 epochs. We further quantify estimator bias and the peak-index MSE–SNR law under phase-quantization (DNL/INL) and VGA nonlinearity, and validate the analysis through Monte-Carlo simulation (2–20 dB). The results confirm the efficacy of neural-network-based calibration for achieving robust and online phased-array calibration.


  • Exploiting Expertise of Non-Expert and Diverse Agents in Social Bandit Learning: A Free Energy Approach

    by Erfan Mirzaei, Seyed Pooya Shariatpanahi, Alireza Tavakoli, Reshad Hosseini, Majid Nili Ahmadabadi

    Personalized AI-based services involve a population of individual reinforcement learning agents. However, most reinforcement learning algorithms focus on harnessing individual learning and fail to leverage the social learning capabilities commonly exhibited by humans and animals. Social learning integrates individual experience with observing others’ behavior, presenting opportunities for improved learning outcomes. In this study, we focus on a social bandit learning scenario where a social agent observes other agents’ actions without knowledge of their rewards. The agents independently pursue their own policy without explicit motivation to teach each other. We propose a free energy-based social bandit learning algorithm over the policy space, where the social agent evaluates others’ expertise levels without resorting to any oracle or social norms. Accordingly, the social agent integrates its direct experiences in the environment and others’ estimated policies. The theoretical convergence of our algorithm to the optimal policy is proven. Empirical evaluations validate the superiority of our social learning method over alternative approaches in various scenarios. Our algorithm strategically identifies the relevant agents, even in the presence of random or suboptimal agents, and skillfully exploits their behavioral information. In addition to societies including expert agents, in the presence of relevant but non-expert agents, our algorithm significantly enhances individual learning performance, where most related methods fail. Importantly, it also maintains logarithmic regret.

  • Open Access

    Better than Maximum Likelihood Estimation of Model-based and Model-free Learning Styles

    by Sadjad Yazdani, Abdol-Hossein Vahabie, Babak Nadjar-Araabi, Majid Nili Ahmadabadi

    Introduction: Various decision-making systems collaborate to shape human behavior. Goal-directed and habitual systems are the two primary systems studied by reinforcement learning (RL), with model-based (MB) and model-free (MF) learning styles, respectively. Human behavior can be viewed as a combination of these two decision-making paradigms, achieved by the weighted sum of ...

  • Open Access

    A Survey On Enhancing Reinforcement Learning in Complex Environments: Insights from Human and LLM Feedback

    by Alireza Rashidi Laleh, Majid Nili Ahmadabadi

    Reinforcement learning (RL) is one of the active fields in machine learning, demonstrating remarkable potential in tackling real-world challenges. Despite its promising prospects, this methodology has encountered with issues and challenges, hindering it from achieving the best performance. In particular, these approaches lack decent performance when navigating environments and solving tasks with large observation space, often resulting in sample-inefficiency and prolonged learning times. This issue, commonly referred to as the curse of dimensionality, complicates decision-making for RL agents, necessitating a careful balance between attention and decision-making. RL agents, when augmented with human or large language models' (LLMs) feedback, may exhibit resilience and adaptability, leading to enhanced performance and accelerated learning. Such feedback, conveyed through various modalities or granularities including natural language, serves as a guide for RL agents, aiding them in discerning relevant environmental cues and optimizing decision-making processes. In this survey paper, we mainly focus on problems of two-folds: firstly, we focus on humans or an LLMs assistance, investigating the ways in which these entities may collaborate with the RL agent in order to foster optimal behavior and expedite learning; secondly, we delve into the research papers dedicated to addressing the intricacies of environments characterized by large observation space.

  • Open Access

    CoCoP: Enhancing Text Classification with LLM through Code Completion Prompt

    by Mohammad Mahdi Mohajeri, Mohammad Javad Dousti, Majid Nili Ahmadabadi

    Text classification is a fundamental task in natural language processing (NLP), and large language models (LLMs) have demonstrated their capability to perform this task across various domains. However, the performance of LLMs heavily depends on the quality of their input prompts. Recent studies have also shown that LLMs exhibit remarkable results in code-related tasks. To leverage the capabilities of LLMs in text classification, we propose the Code Completion Prompt (CoCoP) method, which transforms the text classification problem into a code completion task. CoCoP significantly improves text classification performance across diverse datasets by utilizing LLMs' code-completion capability. For instance, CoCoP enhances the accuracy of the SST2 dataset by more than 20%. Moreover, when CoCoP integrated with LLMs specifically designed for code-related tasks (code models), such as CodeLLaMA, this method demonstrates better or comparable performance to few-shot learning techniques while using only one-tenth of the model size. The source code of our proposed method will be available to the public upon the acceptance of the paper.

  • Open Access

    Exploiting Expertise of Non-Expert and Diverse Agents in Social Bandit Learning: A Free Energy Approach

    by Erfan Mirzaei, Alireza Tavakoli, Seyed Pooya Shariatpanahi, Reshad Hosseini, Majid Nili Ahmadabadi

    Personalized AI-based services involve a population of individual reinforcement learning agents. However, most reinforcement learning algorithms focus on harnessing individual learning and fail to leverage the social learning capabilities commonly exhibited by humans and animals. Social learning integrates individual experience with observing others' behavior, presenting opportunities for improved learning outcomes. In this study, we focus on a social bandit learning scenario where a social agent observes other agents' actions without knowledge of their rewards. The agents independently pursue their own rewards without any explicit motivation for teaching one another. We propose a free energy-based social bandit learning algorithm over policy space, where the social agent evaluates others' expertise levels without resorting to any oracle or social norms. Accordingly, the social agent integrates its direct experiences in the environment and others' estimated policies. The theoretical convergence of our algorithm to the optimal policy is proven. Empirical evaluations validate the superiority of our social learning method over alternative approaches in diverse scenarios. Our algorithm strategically identifies the relevant agents, even in the presence of random or sub-optimal agents, and skillfully exploiters their behavioral information. Extra to societies including expert agents, in the presence of relevant but non-expert agents, our algorithm significantly enhances individual learning performance, where most related methods fail. Importantly, it maintains logarithmic regret.

  • An Online Calibration Method Using Hadamard–Fourier Clustering and Neural Network for Large-Scale Phased Arrays

    by Zahra Sarayloo, Nasser Masoumi, Ehsan Haj-Mirza-Alian, Naimeh Ghafarian, Mohammad Reza Nezhadahmadi, Majid Nili Ahmadabadi, Safieddin Safavi-Naeini

    This paper proposes a novel online calibration method based on clustering for large-scale phased array antennas. The proposed clustering method (called HaF clustering) leverages Hadamard and Fourier transform features, resulting in increased output power variation (suitable for large phased array calibration), noise robustness, and fewer measurements compared to traditional methods. Additionally, it eliminates the need for extra phase measurement instruments as it relies solely on power measurements. Analytical closed-forms are derived to demonstrate the effectiveness of HaF clustering. In this method, the mean phase error ( MPhE ) in each cluster is determined by a combination of Hadamard features and the extended rotational electrical vector field method (eREV). Using a trained multi-layer perceptron neural network and feeding it with each cluster’s MPhEs , the direction of arrival (DOA) error is determined. Subsequently, antenna phase errors are estimated based on the DOA error, and new calibration coefficients are applied to the array. To validate the proposed online calibration method, Monte-Carlo simulations, and experimental measurements were conducted on a 1024-element modular planar phased array receiver within the frequency range of 18 GHz – 21 GHz and an angle of elevation range between -70 to 70 degrees. The simulation and experimental results indicate a mean absolute error (MAE) value of approximately 6 degrees for phase error determination and a DOA estimation error of less than 0.1 degrees using the multi-layer perceptron (MLP). Furthermore, the array can be calibrated with a maximum calibration error of less than 0.1 degrees within a period of 3 milliseconds.


  • Open Access

    Risk-Sensitive Multi-Agent Reinforcement Learning in Network Aggregative Markov Games

    by Hafez Ghaemi, Hamed Kebriaei, Alireza Ramezani Moghaddam, Majid Nili Ahamdabadi

    Classical multi-agent reinforcement learning (MARL) assumes risk neutrality and complete objectivity for agents. However, in settings where agents need to consider or model human economic or social preferences, a notion of risk must be incorporated into the RL optimization problem. This will be of greater importance in MARL where other human or non-human agents are involved, possibly with their own risk-sensitive policies. In this work, we consider risk-sensitive and non-cooperative MARL with cumulative prospect theory (CPT), a non-convex risk measure and a generalization of coherent measures of risk. CPT is capable of explaining loss aversion in humans and their tendency to overestimate/underestimate small/large probabilities. We propose a distributed sampling-based actor-critic (AC) algorithm with CPT risk for network aggregative Markov games (NAMGs), which we call Distributed Nested CPT-AC. Under a set of assumptions, we prove the convergence of the algorithm to a subjective notion of Markov perfect Nash equilibrium in NAMGs. The experimental results show that subjective CPT policies obtained by our algorithm can be different from the risk-neutral ones, and agents with a higher loss aversion are more inclined to socially isolate themselves in an NAMG.