Sriraam Natarajan
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
Sriraam Natarajan has a total of 122 co-authors. Here are the top co-authors by number of shared publications:
- Kristian Kersting
Technical University of Darmstadt
8 shared publications - Sahil Sidheekh
The University of Texas at Dallas
5 shared publications - Siwen Yan
Statistical Artificial Intelligence and Relational Learning Group (StARLinG Lab) University of Texas
5 shared publications - Gautam Kunapuli
Department of Computer Science The University of Texas at Dallas
4 shared publications - Mayukh Das
Erik Jonsson School of Engineering and Computer Science, The University of Texas at Dallas, United States
4 shared publications - Janardhan Rao Doppa
Washington State University
3 shared publications
Latest Publications
- 2 July 2026
- Published by Springer Nature
- p. 501-525
Human-Allied Learning of Probabilistic Models from Relational Data
- 2 July 2026
- Published by Springer Nature
- p. 501-525
Human-Allied Learning of Probabilistic Models from Relational Data
- Open Access
Human-Allied Relational Reinforcement Learning
Reinforcement learning (RL) has experienced a second wind in the past decade. While incredibly successful in images and videos, these systems still operate within the realm of propositional tasks ignoring the inherent structure that exists in the problem. Consequently, relational extensions (RRL) have been developed for such structured problems that allow for effective generalization to arbitrary number of objects. However, they inherently make strong assumptions about the problem structure. We introduce a novel framework that combines RRL with object-centric representation to handle both structured and unstructured data. We enhance learning by allowing the system to actively query the human expert for guidance by explicitly modeling the uncertainty over the policy. Our empirical evaluation demonstrates the effectiveness and efficiency of our proposed approach.
Open AccessHuman-Allied Relational Reinforcement Learning
Reinforcement learning (RL) has experienced a second wind in the past decade. While incredibly successful in images and videos, these systems still operate within the realm of propositional tasks ignoring the inherent structure that exists in the problem. Consequently, relational extensions (RRL) have been developed for such structured problems that allow for effective generalization to arbitrary number of objects. However, they inherently make strong assumptions about the problem structure. We introduce a novel framework that combines RRL with object-centric representation to handle both structured and unstructured data. We enhance learning by allowing the system to actively query the human expert for guidance by explicitly modeling the uncertainty over the policy. Our empirical evaluation demonstrates the effectiveness and efficiency of our proposed approach.
- Open Access
- 15 October 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society
- Vol. 8 (1) , 978-989
IndiCASA: A Dataset and Bias Evaluation Framework for LLMs Using Contrastive Embedding Similarity in the Indian Context
Large Language Models (LLMs) have gained significant traction across critical domains owing to their impressive contextual understanding and generative capabilities. However, their increasing deployment in high stakes applications necessitates rigorous evaluation of embedded biases, particularly in culturally diverse contexts like India where existing embedding-based bias assessment methods often fall short in capturing nuanced stereotypes. We propose an evaluation framework based on a encoder trained using contrastive learning that captures fine-grained bias through embedding similarity. We also introduce a novel dataset - IndiCASA (IndiBias-based Contextually Aligned Stereotypes and Anti-stereotypes) comprising 2,575 human-validated sentences spanning five demographic axes: caste, gender, religion, disability, and socioeconomic status. Our evaluation of multiple open-weight LLMs reveals that all models exhibit some degree of stereotypical bias, with disability related biases being notably persistent, and religion bias generally lower likely due to global debiasing efforts demonstrating the need for fairer model development.
- 15 October 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society
- Vol. 8 (1) , 978-989
Open AccessIndiCASA: A Dataset and Bias Evaluation Framework for LLMs Using Contrastive Embedding Similarity in the Indian Context
Large Language Models (LLMs) have gained significant traction across critical domains owing to their impressive contextual understanding and generative capabilities. However, their increasing deployment in high stakes applications necessitates rigorous evaluation of embedded biases, particularly in culturally diverse contexts like India where existing embedding-based bias assessment methods often fall short in capturing nuanced stereotypes. We propose an evaluation framework based on a encoder trained using contrastive learning that captures fine-grained bias through embedding similarity. We also introduce a novel dataset - IndiCASA (IndiBias-based Contextually Aligned Stereotypes and Anti-stereotypes) comprising 2,575 human-validated sentences spanning five demographic axes: caste, gender, religion, disability, and socioeconomic status. Our evaluation of multiple open-weight LLMs reveals that all models exhibit some degree of stereotypical bias, with disability related biases being notably persistent, and religion bias generally lower likely due to global debiasing efforts demonstrating the need for fairer model development.
- Open Access
IndiCASA: A Dataset and Bias Evaluation Framework in LLMs Using Contrastive Embedding Similarity in the Indian Context
Large Language Models (LLMs) have gained significant traction across critical domains owing to their impressive contextual understanding and generative capabilities. However, their increasing deployment in high stakes applications necessitates rigorous evaluation of embedded biases, particularly in culturally diverse contexts like India where existing embedding-based bias assessment methods often fall short in capturing nuanced stereotypes. We propose an evaluation framework based on a encoder trained using contrastive learning that captures fine-grained bias through embedding similarity. We also introduce a novel dataset - IndiCASA (IndiBias-based Contextually Aligned Stereotypes and Anti-stereotypes) comprising 2,575 human-validated sentences spanning five demographic axes: caste, gender, religion, disability, and socioeconomic status. Our evaluation of multiple open-weight LLMs reveals that all models exhibit some degree of stereotypical bias, with disability related biases being notably persistent, and religion bias generally lower likely due to global debiasing efforts demonstrating the need for fairer model development.
Open AccessIndiCASA: A Dataset and Bias Evaluation Framework in LLMs Using Contrastive Embedding Similarity in the Indian Context
Large Language Models (LLMs) have gained significant traction across critical domains owing to their impressive contextual understanding and generative capabilities. However, their increasing deployment in high stakes applications necessitates rigorous evaluation of embedded biases, particularly in culturally diverse contexts like India where existing embedding-based bias assessment methods often fall short in capturing nuanced stereotypes. We propose an evaluation framework based on a encoder trained using contrastive learning that captures fine-grained bias through embedding similarity. We also introduce a novel dataset - IndiCASA (IndiBias-based Contextually Aligned Stereotypes and Anti-stereotypes) comprising 2,575 human-validated sentences spanning five demographic axes: caste, gender, religion, disability, and socioeconomic status. Our evaluation of multiple open-weight LLMs reveals that all models exhibit some degree of stereotypical bias, with disability related biases being notably persistent, and religion bias generally lower likely due to global debiasing efforts demonstrating the need for fairer model development.
- 22 June 2025
- Published by Springer Nature
- p. 255-260
LLM-Guided Causal Bayesian Network Construction for Pediatric Patients on ECMO
- 22 June 2025
- Published by Springer Nature
- p. 255-260
LLM-Guided Causal Bayesian Network Construction for Pediatric Patients on ECMO
- 18 June 2025
- Published by Springer Nature
- p. 469-481
Scalable Knowledge Graph Construction from Unstructured Text: A Case Study on Artisanal and Small-Scale Gold Mining
- 18 June 2025
- Published by Springer Nature
- p. 469-481
Scalable Knowledge Graph Construction from Unstructured Text: A Case Study on Artisanal and Small-Scale Gold Mining
- Open Access
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (17) , 17779-17787
A Unified Framework for Human-Allied Learning of Probabilistic Circuits
Probabilistic Circuits (PCs) have emerged as an efficient framework for representing and learning complex probability distributions. Nevertheless, the existing body of research on PCs predominantly concentrates on data-driven parameter learning, often neglecting the potential of knowledge-intensive learning, a particular issue in data-scarce/knowledge-rich domains such as healthcare. To bridge this gap, we propose a novel unified framework that can systematically integrate diverse domain knowledge into the parameter learning process of PCs. Experiments on several benchmarks as well as real world datasets show that our proposed framework can both effectively and efficiently leverage domain knowledge to achieve superior performance compared to purely data-driven learning approaches.
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (17) , 17779-17787
Open AccessA Unified Framework for Human-Allied Learning of Probabilistic Circuits
Probabilistic Circuits (PCs) have emerged as an efficient framework for representing and learning complex probability distributions. Nevertheless, the existing body of research on PCs predominantly concentrates on data-driven parameter learning, often neglecting the potential of knowledge-intensive learning, a particular issue in data-scarce/knowledge-rich domains such as healthcare. To bridge this gap, we propose a novel unified framework that can systematically integrate diverse domain knowledge into the parameter learning process of PCs. Experiments on several benchmarks as well as real world datasets show that our proposed framework can both effectively and efficiently leverage domain knowledge to achieve superior performance compared to purely data-driven learning approaches.
- Open Access
Combining Planning and Reinforcement Learning for Solving Relational Multiagent Domains
Multiagent Reinforcement Learning (MARL) poses significant challenges due to the exponential growth of state and action spaces and the non-stationary nature of multiagent environments. This results in notable sample inefficiency and hinders generalization across diverse tasks. The complexity is further pronounced in relational settings, where domain knowledge is crucial but often underutilized by existing MARL algorithms. To overcome these hurdles, we propose integrating relational planners as centralized controllers with efficient state abstractions and reinforcement learning. This approach proves to be sample-efficient and facilitates effective task transfer and generalization.
Open AccessCombining Planning and Reinforcement Learning for Solving Relational Multiagent Domains
Multiagent Reinforcement Learning (MARL) poses significant challenges due to the exponential growth of state and action spaces and the non-stationary nature of multiagent environments. This results in notable sample inefficiency and hinders generalization across diverse tasks. The complexity is further pronounced in relational settings, where domain knowledge is crucial but often underutilized by existing MARL algorithms. To overcome these hurdles, we propose integrating relational planners as centralized controllers with efficient state abstractions and reinforcement learning. This approach proves to be sample-efficient and facilitates effective task transfer and generalization.
- Open Access
- 1 August 2024
- Published by International Joint Conferences on Artificial Intelligence
- Vol. 9, 8186-8194
KnowRA: Knowledge Retrieval Augmented Method for Document-level Relation Extraction with Comprehensive Reasoning Abilities
Document-level relation extraction (Doc-RE) aims to extract relations between entities across multiple sentences. Therefore, Doc-RE requires more comprehensive reasoning abilities like humans, involving complex cross-sentence interactions between entities, contexts, and external general knowledge, compared to the sentence-level RE. However, most existing Doc-RE methods focus on optimizing single reasoning ability, but lack the ability to utilize external knowledge for comprehensive reasoning on long documents. To solve these problems, a knowledge retrieval augmented method, named KnowRA, was proposed with comprehensive reasoning to autonomously determine whether to accept external knowledge to assist Doc-RE. Firstly, we constructed a document graph for semantic encoding and integrated the co-reference resolution model to augment the co-reference reasoning ability. Then, we expanded the document graph into a document knowledge graph by retrieving the external knowledge base for common-sense reasoning and a novel knowledge filtration method was presented to filter out irrelevant knowledge. Finally, we proposed the axis attention mechanism to build direct and indirect associations with intermediary entities for achieving cross-sentence logical reasoning. Extensive experiments conducted on two datasets verified the effectiveness of our method compared to the state-of-the-art baselines. Our code is available at https://anonymous.4open.science/r/KnowRA.
- 1 August 2024
- Published by International Joint Conferences on Artificial Intelligence
- Vol. 9, 8186-8194
Open AccessKnowRA: Knowledge Retrieval Augmented Method for Document-level Relation Extraction with Comprehensive Reasoning Abilities
Document-level relation extraction (Doc-RE) aims to extract relations between entities across multiple sentences. Therefore, Doc-RE requires more comprehensive reasoning abilities like humans, involving complex cross-sentence interactions between entities, contexts, and external general knowledge, compared to the sentence-level RE. However, most existing Doc-RE methods focus on optimizing single reasoning ability, but lack the ability to utilize external knowledge for comprehensive reasoning on long documents. To solve these problems, a knowledge retrieval augmented method, named KnowRA, was proposed with comprehensive reasoning to autonomously determine whether to accept external knowledge to assist Doc-RE. Firstly, we constructed a document graph for semantic encoding and integrated the co-reference resolution model to augment the co-reference reasoning ability. Then, we expanded the document graph into a document knowledge graph by retrieving the external knowledge base for common-sense reasoning and a novel knowledge filtration method was presented to filter out irrelevant knowledge. Finally, we proposed the axis attention mechanism to build direct and indirect associations with intermediary entities for achieving cross-sentence logical reasoning. Extensive experiments conducted on two datasets verified the effectiveness of our method compared to the state-of-the-art baselines. Our code is available at https://anonymous.4open.science/r/KnowRA.
- Open Access
- 24 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 83-91
Utilizing Threat Partitioning for More Practical Network Anomaly Detection
Anomaly-based network intrusion detection would appear on the surface to be ideal for detection of zero-day network threats. Yet in practice, their often unacceptably high false positive rates keep them on the sideline in favor of signature-based methods, which typically detect known threats. We argue that an anomaly-based network intrusion detection system should not only be specialized to a specific class of related threats, but characteristics of the threat class itself should be utilized when designing both the detection system and structuring the network data to use with the system. To this end, we take two common network threat classes, DDoS-as-a-Smokescreen (DaaSS) and SYN flood, and analyze their characteristics for structure that we can use to specialize anomaly detection. We partition these threat classes into known behavior and unknown behavior, leaving the latter open-ended. Through experimentation on multiple datasets, we show that our proposed detection system based on this threat partitioning approach is capable of detecting DaaSS attacks and zero-day SYN flood variants with very low false positive rates, even in the face of concept drift, and can do so without having to collect large amounts of benign network traffic for training.
- 24 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 83-91
Open AccessUtilizing Threat Partitioning for More Practical Network Anomaly Detection
Anomaly-based network intrusion detection would appear on the surface to be ideal for detection of zero-day network threats. Yet in practice, their often unacceptably high false positive rates keep them on the sideline in favor of signature-based methods, which typically detect known threats. We argue that an anomaly-based network intrusion detection system should not only be specialized to a specific class of related threats, but characteristics of the threat class itself should be utilized when designing both the detection system and structuring the network data to use with the system. To this end, we take two common network threat classes, DDoS-as-a-Smokescreen (DaaSS) and SYN flood, and analyze their characteristics for structure that we can use to specialize anomaly detection. We partition these threat classes into known behavior and unknown behavior, leaving the latter open-ended. Through experimentation on multiple datasets, we show that our proposed detection system based on this threat partitioning approach is capable of detecting DaaSS attacks and zero-day SYN flood variants with very low false positive rates, even in the face of concept drift, and can do so without having to collect large amounts of benign network traffic for training.