Nikolaus Salvatore

PhD Student in Computer Science, Rutgers University

Human memory-inspired AI · Large language models · Computational cognitive science

I am a PhD student in Computer Science at Rutgers University-New Brunswick and a member of the Memory Optimization Lab. I study how cognitive principles of human memory can be leveraged to improve the memory capabilities of neural network architectures. My current work connects computational models of human memory with neural sequence models and investigates how aspects of human memory can enhance memory-augmented LLM architectures.

Before Rutgers, I worked on event-based computer vision, neuromorphic computing, spiking neural networks, adaptive sensing, and space-domain awareness. Across these areas, I am interested in how intelligent systems allocate limited representational and computational resources to find relevant information.

I earned an M.E. in Computer Engineering from the University of Pittsburgh and a B.S. in Biological Sciences (minor in CS) from Cornell University. My research background also includes appointments with KBR, the Air Force Research Laboratory, and the NSF Center for Space, High-Performance, and Resilient Computing.

research

Journal Articles & Peer-reviewed Publications

Sequence-to-sequence models with attention mechanistically map to the architecture of human memory search
N. Salvatore and Q. Zhang
Communications Psychology, 3, Article 146, 2025
[Journal] [PubMed]

Parallels between neural machine translation and human memory search: A cognitive modeling approach
N. Salvatore and Q. Zhang
Proceedings of the 46th Annual Conference of the Cognitive Science Society, 2024
[Paper]

Dynamic vision-based satellite detection: A time-based encoding approach with spiking neural networks
N. Salvatore and J. Fletcher
International Conference on Computer Vision Systems, LNCS 14253, 2023
[Paper]

Learned event-based visual perception for improved space object detection
N. Salvatore and J. Fletcher
IEEE/CVF Winter Conference on Applications of Computer Vision, 2022
[Paper] [PDF]

A neuro-inspired approach to intelligent collision avoidance and navigation
N. Salvatore, S. Mian, C. Abidi, and A. D. George
AIAA/IEEE 39th Digital Avionics Systems Conference, 2020
[Paper]

Event-based noise filtration with point-of-interest detection and tracking for space situational awareness
N. Salvatore and A. D. George
2nd International Conference on Intelligent Medicine and Image Processing, 2020
[Paper]

preprint

Lost in the Middle: An Emergent Property from Information Retrieval Demands in LLMs
N. Salvatore, H. Wang, and Q. Zhang
arXiv:2510.10276, 2025
[arXiv]

Projects

current and prior research themes

Information retrieval demands in large language modelsInvestigating whether primacy, recency, and lost-in-the-middle behavior arise as adaptations to mixtures of short-term and long-term retrieval demands during training. The project connects positional bias in LLMs to classic human memory paradigms and structural attention dynamics.Rutgers University · 2025-present
[Preprint]
large language modelsinformation retrievalpositional biasattention dynamics

Human memory and neural sequence modelsDeveloping interpretable neural-network models of human memory search by mapping sequence-to-sequence architectures and attention mechanisms onto context-based cognitive models. This work asks why particular memory architectures emerge and how learning produces human-like recall dynamics.Rutgers University · 2023-present
[Journal article] [Code]
cognitive modelinghuman memoryattentionsequence models

Event-based perception for space-domain awarenessDeveloped learning-based approaches for detecting, classifying, denoising, and tracking faint space objects in sparse asynchronous event-camera data. The work includes hybrid frame/event models, spiking neural networks, and adaptive sensing methods.KBR, AFRL, and NSF SHREC · 2018-2024
[WACV paper] [ICVS paper]
event-based visionneuromorphic computingspiking neural networksspace systems

Talks & Presentations

Presentation at the Context and Episodic Memory Symposium
CEMS, Philadelphia, Pennsylvania · 2026

A neural network model of free recall and its connection to neural machine translation
57th Annual Meeting of the Society for Mathematical Psychology and 22nd International Conference on Cognitive Modeling, Tilburg, Netherlands · 2024
[Program]

Parallels between neural machine translation and human memory search: A cognitive modeling approach
46th Annual Conference of the Cognitive Science Society · 2024
[Paper]

Parallels between neural machine translation and human memory search: A cognitive modeling approach
Context and Episodic Memory Symposium, Philadelphia, Pennsylvania (poster) · 2024
[Program]

Dynamic vision-based satellite detection: A time-based encoding approach with spiking neural networks
International Conference on Computer Vision Systems · 2023

Learned event-based visual perception for improved space object detection
IEEE/CVF Winter Conference on Applications of Computer Vision · 2022

A neuro-inspired approach to intelligent collision avoidance and navigation
AIAA/IEEE 39th Digital Avionics Systems Conference · 2020

Event-based noise filtration with point-of-interest detection and tracking for space situational awareness
2nd International Conference on Intelligent Medicine and Image Processing · 2020

Teaching

Teaching Assistant, Department of Computer Science

Rutgers University-New Brunswick · 2023-present
01:198:211 Computer Architecture
01:198:213 Software Methodology
01:198:440 Introduction to Artificial Intelligence