arXiv subjects

MohammadAmin Fazli

Publications and source records attributed to MohammadAmin Fazli.

At least 19 recordsLinked to original sources

arXiv2026

Hidden Meanings in Plain Sight: RebusBench for Evaluating Cognitive Visual Reasoning

Large Vision-Language Models (LVLMs) have achieved remarkable proficiency in explicit visual recognition, effectively describing what is directly visible in an image. However, a critical cognitive gap emerges when the visual input serves only as a clue rather than the answer. We identify that current models struggle with the complex, multi-step reasoning required to solve problems where information is not explicitly depicted. Successfully solving a rebus puzzle requires a distinct cognitive workflow: the model must extract visual and textual attributes, retrieve linguistic prior knowledge (such as idioms), and perform abstract mapping to synthesize these elements into a meaning that exists outside the pixel space. To evaluate this neurosymbolic capability, we introduce RebusBench, a benchmark of 1,164 puzzles designed to test this specific integration of perception and knowledge. Our evaluation of state-of-the-art models (including Qwen, InternVL, and LLaVA) shows a severe deficiency: performance saturates below 10% Exact Match and 20% semantic accuracy, with no significant improvement observed from model scaling or In-Context Learning (ICL). These findings suggest that while models possess the necessary visual and linguistic components, they lack the cognitive reasoning glue to connect them. Project page available at https://amirkasaei.com/rebusbench/.

cs.CV↗

arXiv2025

From Illusion to Insight: Change-Aware File-Level Software Defect Prediction Using Agentic AI

Much of the reported progress in file-level software defect prediction (SDP) is, in reality, nothing but an illusion of accuracy. Over the last decades, machine learning and deep learning models have reported increasing performance across software versions. However, since most files persist across releases and retain their defect labels, standard evaluation rewards label-persistence bias rather than reasoning about code changes. To address this issue, we reformulate SDP as a change-aware prediction task, in which models reason over code changes of a file within successive project versions, rather than relying on static file snapshots. Building on this formulation, we propose an LLM-driven, change-aware, multi-agent debate framework. Our experiments on multiple PROMISE projects show that traditional models achieve inflated F1, while failing on rare but critical defect-transition cases. In contrast, our change-aware reasoning and multi-agent debate framework yields more balanced performance across evolution subsets and significantly improves sensitivity to defect introductions. These results highlight fundamental flaws in current SDP evaluation practices and emphasize the need for change-aware reasoning in practical defect prediction. The source code is publicly available.

cs.SE↗

arXiv2025

RealDrag: The First Dragging Benchmark with Real Target Image

The evaluation of drag based image editing models is unreliable due to a lack of standardized benchmarks and metrics. This ambiguity stems from inconsistent evaluation protocols and, critically, the absence of datasets containing ground truth target images, making objective comparisons between competing methods difficult. To address this, we introduce \textbf{RealDrag}, the first comprehensive benchmark for point based image editing that includes paired ground truth target images. Our dataset contains over 400 human annotated samples from diverse video sources, providing source/target images, handle/target points, editable region masks, and descriptive captions for both the image and the editing action. We also propose four novel, task specific metrics: Semantical Distance (SeD), Outer Mask Preserving Score (OMPS), Inner Patch Preserving Score (IPPS), and Directional Similarity (DiS). These metrics are designed to quantify pixel level matching fidelity, check preservation of non edited (out of mask) regions, and measure semantic alignment with the desired task. Using this benchmark, we conduct the first large scale systematic analysis of the field, evaluating 17 SOTA models. Our results reveal clear trade offs among current approaches and establish a robust, reproducible baseline to guide future research. Our dataset and evaluation toolkit will be made publicly available.

cs.CV↗

arXiv2025

Evaluating the Evaluators: Metrics for Compositional Text-to-Image Generation

Text-image generation has advanced rapidly, but assessing whether outputs truly capture the objects, attributes, and relations described in prompts remains a central challenge. Evaluation in this space relies heavily on automated metrics, yet these are often adopted by convention or popularity rather than validated against human judgment. Because evaluation and reported progress in the field depend directly on these metrics, it is critical to understand how well they reflect human preferences. To address this, we present a broad study of widely used metrics for compositional text-image evaluation. Our analysis goes beyond simple correlation, examining their behavior across diverse compositional challenges and comparing how different metric families align with human judgments. The results show that no single metric performs consistently across tasks: performance varies with the type of compositional problem. Notably, VQA-based metrics, though popular, are not uniformly superior, while certain embedding-based metrics prove stronger in specific cases. Image-only metrics, as expected, contribute little to compositional evaluation, as they are designed for perceptual quality rather than alignment. These findings underscore the importance of careful and transparent metric selection, both for trustworthy evaluation and for their use as reward models in generation. Project page is available at https://amirkasaei.com/eval-the-evals/ .

cs.CV↗

arXiv2025

Cold-Start Active Preference Learning in Socio-Economic Domains

Active preference learning offers an efficient approach to modeling preferences, but it is hindered by the cold-start problem, which leads to a marked decline in performance when no initial labeled data are available. While cold-start solutions have been proposed for domains such as vision and text, the cold-start problem in active preference learning remains largely unexplored, underscoring the need for practical, effective methods. Drawing inspiration from established practices in social and economic research, the proposed method initiates learning with a self-supervised phase that employs Principal Component Analysis (PCA) to generate initial pseudo-labels. This process produces a \say{warmed-up} model based solely on the data's intrinsic structure, without requiring expert input. The model is then refined through an active learning loop that strategically queries a simulated noisy oracle for labels. Experiments conducted on various socio-economic datasets, including those related to financial credibility, career success rate, and socio-economic status, consistently show that the PCA-driven approach outperforms standard active learning strategies that start without prior information. This work thus provides a computationally efficient and straightforward solution that effectively addresses the cold-start problem.

cs.LG↗

arXiv2025

CARINOX: Inference-time Scaling with Category-Aware Reward-based Initial Noise Optimization and Exploration

Text-to-image diffusion models, such as Stable Diffusion, can produce high-quality and diverse images but often fail to achieve compositional alignment, particularly when prompts describe complex object relationships, attributes, or spatial arrangements. Recent inference-time approaches address this by optimizing or exploring the initial noise under the guidance of reward functions that score text-image alignment without requiring model fine-tuning. While promising, each strategy has intrinsic limitations when used alone: optimization can stall due to poor initialization or unfavorable search trajectories, whereas exploration may require a prohibitively large number of samples to locate a satisfactory output. Our analysis further shows that neither single reward metrics nor ad-hoc combinations reliably capture all aspects of compositionality, leading to weak or inconsistent guidance. To overcome these challenges, we present Category-Aware Reward-based Initial Noise Optimization and EXploration (CARINOX), a unified framework that combines noise optimization and exploration with a principled reward selection procedure grounded in correlation with human judgments. Evaluations on two complementary benchmarks covering diverse compositional challenges show that CARINOX raises average alignment scores by approximately 19% on both T2I-CompBench++ and HRS, consistently outperforming state-of-the-art optimization and exploration-based methods across all major categories, while preserving image quality and diversity. The project page is available at https://amirkasaei.com/carinox/.

cs.CV↗

arXiv2025

Designing DSIC Mechanisms for Data Sharing in the Era of Large Language Models

Training large language models (LLMs) requires vast amounts of high-quality data from institutions that face legal, privacy, and strategic constraints. Existing data procurement methods often rely on unverifiable trust or ignore heterogeneous provider costs. We introduce a mechanism-design framework for truthful, trust-minimized data sharing that ensures dominant-strategy incentive compatibility (DSIC), individual rationality, and weak budget balance, while rewarding data based on both quality and learning utility. We formalize a model where providers privately know their data cost and quality, and value arises solely from the data's contribution to model performance. Based on this, we propose the Quality-Weighted Marginal-Incentive Auction (Q-MIA), which ranks providers using a virtual cost metric and uses Myerson-style payments to ensure DSIC and budget feasibility. To support settings with limited liquidity or long-term incentives, we introduce the Marginal Utility Token (MUT), which allocates future rights based on marginal contributions. We unify these in Mixed-MIA, a hybrid mechanism balancing upfront payments and deferred rewards. All mechanisms support verifiable, privacy-preserving implementation. Theoretically and empirically, they outperform volume-based and trust-based baselines, eliciting higher-quality data under budget constraints while remaining robust to misreporting and collusion. This establishes a principled foundation for sustainable and fair data markets for future LLMs.

cs.GT↗

arXiv2024

A Decision Support System for Stock Selection and Asset Allocation Based on Fundamental Data Analysis

Financial markets are integral to a country's economic success, yet their complex nature raises challenging issues for predicting their behaviors. There is a growing demand for an integrated system that explores the vast and diverse data in financial reports with powerful machine-learning models to analyze financial markets and suggest appropriate investment strategies. This research provides an end-to-end decision support system (DSS) that pervasively covers the stages of gathering, cleaning, and modeling the stock's financial and fundamental data alongside the country's macroeconomic conditions. Analyzing and modeling the fundamental data of securities is a noteworthy method that, despite its greater power, has been used by fewer researchers due to its more complex and challenging issues. By precisely analyzing securities' fundamental data, the proposed system assists investors in predicting stock future prices and allocating assets in major financial markets: stock, bond, and commodity. The most notable contributions and innovations of this research are: (1) Developing a robust predictive model for mid- to long-term stock returns, tailored for investors rather than traders, (2) The proposed DSS considers a diverse set of features relating to the economic conditions of the company, including fundamental data, stock trading characteristics, and macro-economic attributes to enhance predictive accuracy, (3) Evaluating the DSS performance on the Tehran Stock Exchange that has specific characteristics of small to medium-sized economies with high inflation rates and showing the superiority to novel researches, and (4) Empowering the DSS to generate different asset allocation strategies in various economic situations by simulating expert investor decision-making.

q-fin.ST↗

arXiv2024

A Comprehensive Survey of Convolutions in Deep Learning: Applications, Challenges, and Future Trends

In today's digital age, Convolutional Neural Networks (CNNs), a subset of Deep Learning (DL), are widely used for various computer vision tasks such as image classification, object detection, and image segmentation. There are numerous types of CNNs designed to meet specific needs and requirements, including 1D, 2D, and 3D CNNs, as well as dilated, grouped, attention, depthwise convolutions, and NAS, among others. Each type of CNN has its unique structure and characteristics, making it suitable for specific tasks. It's crucial to gain a thorough understanding and perform a comparative analysis of these different CNN types to understand their strengths and weaknesses. Furthermore, studying the performance, limitations, and practical applications of each type of CNN can aid in the development of new and improved architectures in the future. We also dive into the platforms and frameworks that researchers utilize for their research or development from various perspectives. Additionally, we explore the main research fields of CNN like 6D vision, generative models, and meta-learning. This survey paper provides a comprehensive examination and comparison of various CNN architectures, highlighting their architectural differences and emphasizing their respective advantages, disadvantages, applications, challenges, and future trends.

cs.LG↗

arXiv2023

Effect of Choosing Loss Function when Using T-batching for Representation Learning on Dynamic Networks

Representation learning methods have revolutionized machine learning on networks by converting discrete network structures into continuous domains. However, dynamic networks that evolve over time pose new challenges. To address this, dynamic representation learning methods have gained attention, offering benefits like reduced learning time and improved accuracy by utilizing temporal information. T-batching is a valuable technique for training dynamic network models that reduces training time while preserving vital conditions for accurate modeling. However, we have identified a limitation in the training loss function used with t-batching. Through mathematical analysis, we propose two alternative loss functions that overcome these issues, resulting in enhanced training performance. We extensively evaluate the proposed loss functions on synthetic and real-world dynamic networks. The results consistently demonstrate superior performance compared to the original loss function. Notably, in a real-world network characterized by diverse user interaction histories, the proposed loss functions achieved more than 26.9% enhancement in Mean Reciprocal Rank (MRR) and more than 11.8% improvement in Recall@10. These findings underscore the efficacy of the proposed loss functions in dynamic network modeling.

cs.LG↗

arXiv2022

A Novel Experts Advice Aggregation Framework Using Deep Reinforcement Learning for Portfolio Management

Solving portfolio management problems using deep reinforcement learning has been getting much attention in finance for a few years. We have proposed a new method using experts signals and historical price data to feed into our reinforcement learning framework. Although experts signals have been used in previous works in the field of finance, as far as we know, it is the first time this method, in tandem with deep RL, is used to solve the financial portfolio management problem. Our proposed framework consists of a convolutional network for aggregating signals, another convolutional network for historical price data, and a vanilla network. We used the Proximal Policy Optimization algorithm as the agent to process the reward and take action in the environment. The results suggested that, on average, our framework could gain 90 percent of the profit earned by the best expert.

q-fin.CP↗

arXiv2021

RPS: Portfolio Asset Selection using Graph based Representation Learning

Portfolio optimization is one of the essential fields of focus in finance. There has been an increasing demand for novel computational methods in this area to compute portfolios with better returns and lower risks in recent years. We present a novel computational method called Representation Portfolio Selection (RPS) by redefining the distance matrix of financial assets using Representation Learning and Clustering algorithms for portfolio selection to increase diversification. RPS proposes a heuristic for getting closer to the optimal subset of assets. Using empirical results in this paper, we demonstrate that widely used portfolio optimization algorithms, such as MVO, CLA, and HRP, can benefit from our asset subset selection.

q-fin.PM↗

arXiv2021

Under the Skin of Foundation NFT Auctions

Non Fungible Tokens (NFTs) have gained a solid foothold within the crypto community, and substantial amounts of money have been allocated to their trades. In this paper, we studied one of the most prominent marketplaces dedicated to NFT auctions and trades, Foundation. We analyzed the activities on Foundation and identified several intriguing underlying dynamics that occur on this platform. Moreover, We performed social network analysis on a graph that we had created based on transferred NFTs on Foundation, and then described the characteristics of this graph. Lastly, We built a neural network-based similarity model for retrieving and clustering similar NFTs. We also showed that for most NFTs, their performances in auctions were comparable with the auction performance of other NFTs in their cluster.

cs.SI↗

arXiv2021

Building Stable Off-chain Payment Networks

Payment channel is a protocol which allows cryptocurrency users to route multiple transactions through network without committing them to the main blockchain network (mainnet). This ability makes them the most prominent solution to blockchains' scalability problem. Each modification of payment channels requires a transaction on the mainnet and therefore, big transaction fees. In this paper, we assume that a set of payment transactions are given (batch or online) and we study the problem of scheduling modificiations on payment channels to route all of the transactions with minimum modification cost. We investigate two cost models for aforementioned problem: the step cost function in which every channel modification has a constant cost and the linear cost function in which modification costs are proportional to the amount of change. For the step cost function model, we prove impossibility results for both batch and online case. Moreover, some heuristic methods for the batch case are presented and compared. For the linear cost we propose a polynomial time algorithm using linear programming for the batch case.

cs.GT↗

arXiv2019

Governance of Social Welfare in Networked Markets

This paper aims to investigate how a central authority (e.g. a government) can increase social welfare in a network of markets and firms. In these networks, modeled using a bipartite graph, firms compete with each other \textit{à la} Cournot. Each firm can supply homogeneous goods in markets which it has access to. The central authority may take different policies for its aim. In this paper, we assume that the government has a budget by which it can supply some goods and inject them into various markets. We discuss how the central authority can best allocate its budget for the distribution of goods to maximize social welfare. We show that the solution is highly dependent on the structure of the network. Then, using the network's structural features, we present a heuristic algorithm for our target problem. Finally, we compare the performance of our algorithm with other heuristics with experimentation on real datasets.

cs.GT↗

arXiv2019

Reactive Liquid: Optimized Liquid Architecture for Elastic and Resilient Distributed Data Processing

Today's most prominent IT companies are built on the extraction of insight from data, and data processing has become crucial in data-intensive businesses. Nevertheless, the size of data which should be processed is growing significantly fast. The pace of the data growing has changed the nature of data processing. Today, data-intensive industries demand highly scalable and fault tolerant data processing architectures which can handle the massive amount of data. In this paper, we presented a distributed architecture for elastic and resilient data processing based on the Liquid which is a nearline and offline big data architecture. We used the Reactive Manifesto to design the architecture highly reactive to workload changes and failures. We evaluate our architecture by drawing some numerical comparisons between our architecture prototype and the Liquid prototype. The performed evaluation shows that our architecture can be more scalable against workload and more resilient against failures than the Liquid architecture is.

cs.DC↗

arXiv2016

Maximizing Non-monotone Submodular Set Functions Subject to Different Constraints: Combined Algorithms

We study the problem of maximizing constrained non-monotone submodular functions and provide approximation algorithms that improve existing algorithms in terms of either the approximation factor or simplicity. Our algorithms combine existing local search and greedy based algorithms. Different constraints that we study are exact cardinality and multiple knapsack constraints. For the multiple-knapsack constraints we achieve a $(0.25-2ε)$-factor algorithm. We also show, as our main contribution, how to use the continuous greedy process for non-monotone functions and, as a result, obtain a $0.13$-factor approximation algorithm for maximization over any solvable down-monotone polytope. The continuous greedy process has been previously used for maximizing smooth monotone submodular function over a down-monotone polytope \cite{CCPV08}. This implies a 0.13-approximation for several discrete problems, such as maximizing a non-negative submodular function subject to a matroid constraint and/or multiple knapsack constraints.

cs.DS↗

arXiv2015

Team Selection For Prediction Tasks

Given a random variable $O \in \mathbb{R}$ and a set of experts $E$, we describe a method for finding a subset of experts $S \subseteq E$ whose aggregated opinion best predicts the outcome of $O$. Therefore, the problem can be regarded as a team formation for performing a prediction task. We show that in case of aggregating experts' opinions by simple averaging, finding the best team (the team with the lowest total error during past $k$ turns) can be modeled with an integer quadratic programming and we prove its NP-hardness whereas its relaxation is solvable in polynomial time. Finally, we do an experimental comparison between different rounding and greedy heuristics and show that our suggested tabu search works effectively. Keywords: Team Selection, Information Aggregation, Opinion Pooling, Quadratic Programming, NP-Hard

cs.DS↗