arXiv subjects

Ashkan Golgoon

Publications and source records attributed to Ashkan Golgoon.

8 recordsLinked to original sources

arXiv2025

Comparative Performance of Collaborative Bandit Algorithms: Effect of Sparsity and Exploration Intensity

This paper offers a comprehensive analysis of collaborative bandit algorithms and provides a thorough comparison of their performance. Collaborative bandits aim to improve the performance of contextual bandits by introducing relationships between arms (or items), allowing effective propagation of information. Collaboration among arms allows the feedback obtained through a single user (item) to be shared across related users (items). Introducing collaboration also alleviates the cold user (item) problem, i.e., lack of historical information when a new user (item) arriving to the platform with no prior record of interactions. In the context of modeling the relationships between arms (items), there are two main approaches: Hard and soft clustering. We call approaches that model the relationship between arms in an \textit{absolute} manner as hard clustering, i.e., the relationship is binary. Soft clustering relaxes membership constraints, allowing \textit{fuzzy} assignment. Focusing on the latter, we provide extensive experiments on the state-of-the-art collaborative contextual bandit algorithms and investigate the effect of sparsity and how the exploration intensity acts as a correction mechanism. Our numerical experiments demonstrate that controlling for sparsity in collaboration improves data efficiency and performance as it better informs learning. Meanwhile, increasing the exploration intensity acts as a correction because it effectively reduces variance due to potentially misspecified relationships among users. We observe that this misspecification is further remedied by introducing latent factors, and thus, increasing the dimensionality of the bandit parameters.

cs.LG↗

arXiv2025

Agentic AI Systems Applied to tasks in Financial Services: Modeling and model risk management crews

The advent of large language models has ushered in a new era of agentic systems, where artificial intelligence programs exhibit remarkable autonomous decision-making capabilities across diverse domains. This paper explores agentic system workflows in the financial services industry. In particular, we build agentic crews with human-in-the-loop module that can effectively collaborate to perform complex modeling and model risk management (MRM) tasks. The modeling crew consists of a judge agent and multiple agents who perform specific tasks such as exploratory data analysis, feature engineering, model selection/hyperparameter tuning, model training, model evaluation, and writing documentation. The MRM crew consists of a judge agent along with specialized agents who perform tasks such as checking compliance of modeling documentation, model replication, conceptual soundness, analysis of outcomes, and writing documentation. We demonstrate the effectiveness and robustness of modeling and MRM crews by presenting a series of numerical examples applied to credit card fraud detection, credit card approval, and portfolio credit risk modeling datasets.

cs.AI↗

arXiv2024

MBExplainer: Multilevel bandit-based explanations for downstream models with augmented graph embeddings

In many industrial applications, it is common that the graph embeddings generated from training GNNs are used in an ensemble model where the embeddings are combined with other tabular features (e.g., original node or edge features) in a downstream ML task. The tabular features may even arise naturally if, e.g., one tries to build a graph such that some of the node or edge features are stored in a tabular format. Here we address the problem of explaining the output of such ensemble models for which the input features consist of learned neural graph embeddings combined with additional tabular features. We propose MBExplainer, a model-agnostic explanation approach for downstream models with augmented graph embeddings. MBExplainer returns a human-legible triple as an explanation for an instance prediction of the whole pipeline consisting of three components: a subgraph with the highest importance, the topmost important nodal features, and the topmost important augmented downstream features. A game-theoretic formulation is used to take the contributions of each component and their interactions into account by assigning three Shapley values corresponding to their own specific games. Finding the explanation requires an efficient search through the corresponding local search spaces corresponding to each component. MBExplainer applies a novel multilevel search algorithm that enables simultaneous pruning of local search spaces in a computationally tractable way. In particular, three interweaved Monte Carlo Tree Search are utilized to iteratively prune the local search spaces. MBExplainer also includes a global search algorithm that uses contextual bandits to efficiently allocate pruning budget among the local search spaces. We show the effectiveness of MBExplainer by presenting a set of comprehensive numerical examples on multiple public graph datasets for both node and graph classification tasks.

cs.LG↗

arXiv2024

Mechanistic interpretability of large language models with applications to the financial services industry

Large Language Models such as GPTs (Generative Pre-trained Transformers) exhibit remarkable capabilities across a broad spectrum of applications. Nevertheless, due to their intrinsic complexity, these models present substantial challenges in interpreting their internal decision-making processes. This lack of transparency poses critical challenges when it comes to their adaptation by financial institutions, where concerns and accountability regarding bias, fairness, and reliability are of paramount importance. Mechanistic interpretability aims at reverse engineering complex AI models such as transformers. In this paper, we are pioneering the use of mechanistic interpretability to shed some light on the inner workings of large language models for use in financial services applications. We offer several examples of how algorithmic tasks can be designed for compliance monitoring purposes. In particular, we investigate GPT-2 Small's attention pattern when prompted to identify potential violation of Fair Lending laws. Using direct logit attribution, we study the contributions of each layer and its corresponding attention heads to the logit difference in the residual stream. Finally, we design clean and corrupted prompts and use activation patching as a causal intervention method to localize our task completion components further. We observe that the (positive) heads $10.2$ (head $2$, layer $10$), $10.7$, and $11.3$, as well as the (negative) heads $9.6$ and $10.6$ play a significant role in the task completion.

cs.LG↗

arXiv2021

Elastodynamic Transformation Cloaking for Non-Centrosymmetric Gradient Solids

In this paper we investigate the possibility of elastodynamic transformation cloaking in bodies made of non-centrosymmetric gradient solids. The goal of transformation cloaking is to hide a hole from elastic disturbances in the sense that the mechanical response of a homogeneous and isotropic body with a hole covered by a cloak would be identical to that of the corresponding homogeneous and isotropic body outside the cloak. It is known that in the case of centrosymmetric gradient solids the balance of angular momentum is the obstruction to transformation cloaking. We will show that this is the case for non-centrosymmetric gradient solids as well.

cond-mat.soft↗

arXiv2020

Transformation Cloaking in Elastic Plates

In this paper we formulate the problem of elastodynamic transformation cloaking for Kirchhoff-Love plates and elastic plates with both the in-plane and out-of-plane displacements. A cloaking transformation maps the boundary-value problem of an isotropic and homogeneous elastic plate (virtual problem) to that of an anisotropic and inhomogeneous elastic plate with a hole surrounded by a cloak that is to be designed (physical problem). For Kirchhoff-Love plates, the (out-of-plane) governing equations of the virtual plate is transformed to those of the physical plate up to an unknown scalar field. In doing so, one finds the initial stress and the initial tangential body force for the physical plate, along with a set of constraints that we call cloaking compatibility equations. These constraints involve the cloaking transformation, the unknown scalar field, and the elastic constants of the virtual plate. It is noted that the cloaking map needs to satisfy certain conditions on the outer boundary of the cloak and the surface of the hole. In particular, the cloaking map needs to fix the outer boundary of the cloak up to the third order. Assuming a generic radial cloaking map, we show that cloaking a circular hole in Kirchoff-Love plates is not possible; the cloaking compatibility equations and the boundary conditions that the cloaking map needs to satisfy are the obstruction to cloaking. Next, relaxing the pure bending assumption, the transformation cloaking problem of an elastic plate in the presence of in-plane and out-of-plane displacements is formulated. In this case, there are two sets of governing equations that need to be simultaneously transformed under the cloaking map. We show that cloaking a circular hole is not possible for a general radial cloaking map; the cloaking compatibility equations and the boundary conditions that the cloaking map needs to satisfy obstruct cloaking.

cond-mat.soft↗

arXiv2018

Nonlinear and Linear Elastodynamics Transformation Cloaking

We formulate the problems of nonlinear and linear elastodynamics transformation cloaking in a geometric framework. It is noted that a cloaking transformation is neither a spatial nor a referential change of frame (coordinates); a cloaking transformation maps the boundary-value problem of an isotropic and homogeneous elastic body (virtual problem) to that of an anisotropic and inhomogeneous elastic body with a hole surrounded by a cloak that is to be designed (physical problem). The virtual body has a desired mechanical response while the physical body is designed to mimic the same response outside the cloak using a cloaking transformation. We show that nonlinear elastodynamics transformation cloaking is not possible while nonlinear elastostatics cloaking may be possible for special deformations, e.g., radial deformations in a body with either a cylindrical or a spherical cavity. For linear elastodynamics, in line with the previous observations in the literature, we show that the elastic constants in the cloak are not fully symmetric; they do not possess the minor symmetries. We prove that transformation cloaking is not possible regardless of the shape of the hole and the cloak. We next show that linear elastodynamics transformation cloaking cannot be achieved for gradient elastic solids either; similar to classical linear elasticity the balance of angular momentum is the obstruction to transformation cloaking. We finally prove that transformation cloaking is not possible for linear elastic generalized Cosserat solids in 2D for any shape of the hole and the cloak. Particularly, in 2D, transformation cloaking cannot be achieved in linear Cosserat elasticity. We also show that transformation cloaking for a spherical cavity covered by a spherical cloak is not possible in the setting of linear elastic generalized Cosserat solids. We conjecture that this result holds for a cavity of any shape.

cond-mat.soft↗

arXiv2018

Line and Point Defects in Nonlinear Anisotropic Solids

In this paper, we present some analytical solutions for the stress fields of nonlinear anisotropic solids with distributed line and point defects. In particular, we determine the stress fields of i) a parallel cylindrically-symmetric distribution of screw dislocations in infinite orthotropic and monoclinic media, ii) a cylindrically-symmetric distribution of parallel wedge disclinations in an infinite orthotropic medium, iii) a distribution of edge dislocations in an orthotropic medium, and iv) a spherically-symmetric distribution of point defects in a transversely isotropic spherical ball.

cond-mat.soft↗