Jiazhen Hong
AI Research Scientist @ EMOTIV · Ph.D. in Computer Engineering @ Rutgers University
Completed his Ph.D. with Prof. Laleh Najafizadeh at the Integrated Systems & NeuroImaging Laboratory. Now an AI Research Scientist at EMOTIV.
Research focuses on foundation models, LLMs, multimodal learning, and scalable machine learning systems for biomedical and time-series data, spanning distributed training, self-supervised representation learning, local model deployment, and healthcare AI applications.
Research Background
Jiazhen Hong is an AI Research Scientist at EMOTIV and holds a Ph.D. in Computer Engineering from Rutgers University. His research focuses on foundation models, large language models, multimodal learning, and scalable AI systems for biomedical and time-series data. At EMOTIV, he works on large-scale neural foundation models, distributed GPU training, long-context modeling, and transferable representation learning. His broader research spans healthcare AI, assistive technologies, human-centered AI, and locally deployed LLM-integrated systems. He has published in Scientific Reports, IEEE TKDE, IEEE TBME, NeurIPS, ICDM, EMBC, ISBI, and NER. Before Rutgers, he conducted machine learning and biostatistics research for early cancer detection at Harvard Medical School.
Recent Highlights
-
2026.10.03
Paper SAMBA accepted to the NeurIPS 2026 Workshop on Symmetry and Geometry in Neural Representations (NeurReps), Proceedings Track.
-
2026.09.28
Career Joined EMOTIV as an AI Research Scientist, working on large-scale neural foundation models, distributed GPU training, and scalable AI systems for biomedical and time-series data.
-
2026.05.15
Award Selected as a finalist in the EMBC 2026 Student Paper Competition, 2026 48th Annual International Conference of Engineering in Medicine and Biology in Toronto, Canada.
-
2026.04.15
Paper ChatBCI-Assist accepted by IEEE Transactions on Biomedical Engineering (TBME).
-
2026.04.15
Paper Two papers accepted at IEEE EMBC 2026.
-
2026.01.06
Award Research Excellence Award, 2025-2026 ECE Development and Research Excellence Award.
-
2025.10.23
Paper ChatBCI accepted by Scientific Reports (Nature Portfolio).
-
2025.09.14
Paper A Geometric Approach to k-means accepted by IEEE Transactions on Knowledge and Data Engineering (TKDE).
📢 Journal & Conference Papers
- 2026.10.03NeurIPS 2026 SAMBA: Toward an EEG Foundation Model via Differential Mamba and Coordinate-Driven Spatial Embedding accepted to the NeurReps Proceedings Track at NeurIPS 2026.
- 2026.04.15TBME ChatBCI-Assist: An Intent-Based P300 Speller with A Locally-Deployed LLM and Adaptive Stopping Strategy Enabling Record Online Spelling Performance accepted by IEEE Transactions on Biomedical Engineering.
- 2026.04.15EMBC 2026 Improving Spelling Performance of LLM-based P300 Speller BCIs using Circular Nonadjacent Flashing and Cross-Validated Calibration accepted at EMBC 2026.
- 2026.04.15EMBC 2026 Llama-4-ALS: A Locally Deployable LLM, Tuned with LoRA and Dynamic Hierarchical Data Augmentation, for Speller-Based AAC accepted at EMBC 2026.
- 2026.01.13ISBI 2026 Neural Dynamics of Time Perception Under Acute Pain: An Auditory Temporal Bisection Study accepted at IEEE ISBI 2026.
- 2025.10.23Sci. Reports ChatBCI, A P300 Speller BCI with Context-Driven Word Prediction Leveraging Large Language Models accepted by Scientific Reports (Nature Portfolio).
- 2025.09.22NeurIPS W SpellerSSL: Self-Supervised Learning with P300 Aggregation for Speller BCIs accepted at NeurIPS Data on the Brain & Mind Workshop.
- 2025.09.20ICDM W An Efficient Self-Supervised Framework for Long-Sequence EEG Modeling accepted at IEEE ICDM AI for Time Series Workshop.
- 2025.09.14TKDE A Geometric Approach to k-means accepted by IEEE Transactions on Knowledge and Data Engineering.
- 2025.08.04NER 2025 Enhancing Typing Speed in LLM-based P300 Speller BCIs Using A New Data-Driven Dynamic Stopping Strategy accepted at IEEE NER 2025.
- 2025.08.04NER 2025 Longest Sustained Cortical Activity Duration in Temporal Bisection Task Aligns with Geometric Mean for Log-Spaced Probes accepted at IEEE NER 2025.
- 2025.04.08EMBC 2025 ChatBCI4ALS: A High-Performance, LLM-Driven Intent-Based BCI Communication System for Individuals with ALS accepted at EMBC 2025.
- 2025.01.02ISBI 2025 TopoEEG: a TimeSformer-Based Topographic Image Representation Method for Early Single-Trial Detection of P300 accepted at IEEE ISBI 2025.
- 2024.09.20IEEE Brain ChatBCI: A Fast P300 Speller Brain-Computer Interface Incorporating Generative AI-Based Word Prediction accepted at 2024 IEEE Brain Discovery & Neurotechnology Workshop.
- 2024.07.29Asilomar P3T: A Transformer Model for Enhancing Character Recognition Rates in P300 Speller Systems accepted at Asilomar ACSSC 2024.
🏆 Awards & Honors
- 2026.01.06Award Selected as a finalist in the EMBC 2026 Student Paper Competition, 2026 48th Annual International Conference of Engineering in Medicine and Biology in Toronto, Canada.
- 2026.01.06Award Research Excellence Award, 2025–2026 Cycle for ECE Development and Research Excellence Award.
- 2025.03.04Award Travel Award, 2024–2025 Research & Conference Travel Award by School of Graduate Studies (SGS).
- 2025.02.28Award Travel Award, 2024–2025 Rutgers Brain Health Institute Trainee Travel Award.
- 2024.09.20Award Travel Award, 2024 IEEE Brain Discovery & Neurotechnology Workshop.
- 2024.05.10Award Best TA Award, Rutgers ECE Department, 2023 Fall.
🎉 Career Milestones
[IEEE TKDE 2025] IEEE Transactions on Knowledge and Data Engineering
We introduce a geometric framework for k-means that characterizes the structure of local optima in clustering.
The approach reveals the geometric relationships between solutions and provides theoretical guarantees for cluster detection under diverse data distributions.
[Nature Portfolio 2026] Scientific Reports
We introduce ChatBCI, a P300 speller that integrates large language models to provide context-driven word prediction during EEG-based spelling.
The system significantly reduces keystrokes and writing time (-53% and -62%) while increasing information transfer rate by 229% compared to conventional letter-by-letter BCI spellers.
[IEEE TBME Special Issue] IEEE Transactions on Biomedical Engineering
We introduce ChatBCI-Assist, an intent-based P300 speller that integrates a locally-deployed large language model with an adaptive stopping strategy for real-time EEG-based communication.
The system achieves record-level online performance, reaching 30.7 characters/min (105.2 bits/min ITR) and 147.1 bits/min semantic information transfer rate, enabling faster and more natural sentence composition in practical BCI communication.
[NeurIPS 2025] The Thirty-ninth Annual Conference on Neural Information Processing Systems
We propose SpellerSSL, a self-supervised learning framework that combines P300 aggregation with cross-domain EEG pretraining to learn generalizable representations for speller BCIs.
The method achieves 94% character recognition with only 7 repetitions, reaches 21.86 bits/min information transfer rate, and reduces calibration data requirements by 60%.
[IEEE ICDM 2025] IEEE International Conference on Data Mining
We introduce EEGM2, a self-supervised framework for long-sequence EEG modeling that integrates a U-shaped encoder–decoder with the Mamba-2 state-space architecture to achieve linear computational complexity and efficient long-range dependency modeling.
The approach improves representation learning for EEG signals, enabling scalable training, faster inference, and strong cross-subject generalization across diverse EEG datasets.
[IEEE NER 2025] IEEE EMBS 12th Annual International Conference on Neural Engineering
We introduce a novel data-driven dynamic stopping strategy for LLM-based P300 speller BCIs, which adaptively determines when to finalize character selection based on real-time EEG decoding.
The method achieves over 260% improvement in information transfer rate compared to traditional fixed-repetition approaches.
[IEEE EMBC 2025] 47th Annual International Conference of the IEEE Engineering in Medicine and Biology Society
We present ChatBCI4ALS, the first intent-based P300 BCI communication system tailored for individuals with ALS.
Online experiments achieve up to 42.16 characters/min typing speed and 128.85 bits/min information transfer rate, significantly advancing real-time BCI communication for assistive applications.
[IEEE ISBI 2025] IEEE International Symposium on Biomedical Imaging
We introduce TopoEEG, an image-based framework that generates sequences of topographic maps from EEG data, preserving both temporal and spatial information for decoding neural dynamics.
Leveraging a TimeSformer backbone, the method enables accurate single-trial early detection of P300 responses.
[Asilomar ACSSC 2024] 58th Asilomar Conference on Signals, Systems, and Computers
We propose P3T (P300-Transformer), a novel single-trial P300 detector and transfer model designed to optimize the information transfer rate in P300-BCI speller systems while maintaining a high character recognition rate.
[IEEE EMBC 2022] 44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society
We propose a deep learning framework utilizing dynamic channel selection for early classification of left and right hand motor imagery tasks, reducing data dimensionality and accelerating real-time BCI decoding.
Selected publications. Bold indicates my name.
2026
[1]
J. Hong, G. Mackellar, and S. Ghane.
SAMBA: Toward an EEG Foundation Model via Differential Mamba and Coordinate-Driven Spatial Embedding.
NeurIPS 2026, NeurReps Proceedings Track.
Workshop
[2]
J. Hong, P. Rao, W. Wang, and L. Najafizadeh.
ChatBCI-Assist: An Intent-Based P300 Speller with a Locally-Deployed LLM and Adaptive Stopping Strategy Enabling Record Online Spelling Performance.
IEEE Transactions on Biomedical Engineering (TBME), 2026.
Journal
IEEE
[3]
J. Hong and L. Najafizadeh.
Improving Spelling Performance of LLM-based P300 Speller BCIs using Circular Nonadjacent Flashing and Cross-Validated Calibration.
IEEE EMBC, 2026.
Conference
[4]
P. Rao, J. Hong, W. Wang, and L. Najafizadeh.
Llama-4-ALS: A Locally Deployable LLM, Tuned with LoRA and Dynamic Hierarchical Data Augmentation, for Speller-Based AAC.
IEEE EMBC, 2026.
Conference
[5]
J. Hong, W. Qian, Y. Chen, and Y. Zhang.
A Geometric Approach to k-means.
IEEE Transactions on Knowledge and Data Engineering (TKDE), 38, 1442–1453, 2026.
Journal
arXiv
IEEE
[6]
A. Subramanian, S. Mai, J. Hong, and L. Najafizadeh.
Neural Dynamics of Time Perception Under Acute Pain: An Auditory Temporal Bisection Study.
IEEE ISBI, 2026.
Conference
[7]
J. Hong, W. Wang, and L. Najafizadeh.
ChatBCI: A P300 Speller BCI with Context-Driven Word Prediction Leveraging Large Language Models, from Concept to Evaluation.
Scientific Reports (Nature Portfolio), 16, 6379, 2026.
Journal
arXiv
Nature
2025
[8]
J. Hong, G. Mackellar, and S. Ghane.
SpellerSSL: Self-Supervised Learning with P300 Aggregation for Speller BCIs.
NeurIPS 2025, Data on the Brain & Mind Workshop.
Workshop
arXiv
[9]
J. Hong, G. Mackellar, and S. Ghane.
An Efficient Self-Supervised Framework for Long-Sequence EEG Modeling.
IEEE ICDM 2025, AI for Time Series Workshop.
Workshop
arXiv
[10]
J. Hong and L. Najafizadeh.
Enhancing Typing Speed in LLM-based P300 Speller BCIs Using a New Data-Driven Dynamic Stopping Strategy.
IEEE EMBS NER, 2025.
Conference
[11]
S. Mai, J. Hong, T. Shors, and L. Najafizadeh.
Longest Sustained Cortical Activity Duration in Temporal Bisection Task Aligns with Geometric Mean for Log-Spaced Probes.
IEEE EMBS NER, 2025.
Conference
[12]
J. Hong, P. Rao, W. Wang, S. Chen, and L. Najafizadeh.
ChatBCI4ALS: A High-Performance, LLM-Driven Intent-Based BCI Communication System for Individuals with ALS.
IEEE EMBC, 2025.
Conference
[13]
J. Hong and L. Najafizadeh.
TopoEEG: a TimeSformer-Based Topographic Image Representation Method for Single-Trial Early Detection of P300.
IEEE ISBI, 2025.
Conference
2024
[14]
J. Hong, W. Wang, and L. Najafizadeh.
ChatBCI: A Fast P300 Speller BCI Incorporating Generative AI-Based Word Prediction.
IEEE Brain Discovery and Neurotechnology Workshop, 2024.
Workshop
Poster
[15]
J. Hong and L. Najafizadeh.
P3T: A Transformer Model for Enhancing Character Recognition Rates in P300 Speller Systems.
Asilomar ACSSC, 2024, pp. 514–518.
Conference
Poster
2022
[16]
J. Hong, F. Shamsi, and L. Najafizadeh.
A Deep Learning Framework Based on Dynamic Channel Selection for Early Classification of Left and Right Hand Motor Imagery Tasks.
IEEE EMBC, 2022, Glasgow, Scotland, pp. 3550–3553.
Conference
IEEE
Poster
Journal papers with expanded details.
Conference and workshop papers with expanded details.
[1]
Workshop
SAMBA: Toward an EEG Foundation Model via Differential Mamba and Coordinate-Driven Spatial Embedding
[2]
Conference
Improving Spelling Performance of LLM-based P300 Speller BCIs using Circular Nonadjacent Flashing and Cross-Validated Calibration
IEEE EMBC, 2026
[3]
Conference
Llama-4-ALS: A Locally Deployable LLM, Tuned with LoRA and Dynamic Hierarchical Data Augmentation, for Speller-Based AAC
IEEE EMBC, 2026
[4]
Conference
Neural Dynamics of Time Perception Under Acute Pain: An Auditory Temporal Bisection Study
IEEE ISBI, 2026
[5]
[6]
Workshop
IEEE ICDM 2025, AI for Time Series Workshop
[7]
Conference
Enhancing Typing Speed in LLM-based P300 Speller BCIs Using a New Data-Driven Dynamic Stopping Strategy
IEEE EMBS NER, 2025
[8]
Conference
Longest Sustained Cortical Activity Duration in Temporal Bisection Task Aligns with Geometric Mean for Log-Spaced Probes
IEEE EMBS NER, 2025
[9]
Conference
ChatBCI4ALS: A High-Performance, LLM-Driven Intent-Based BCI Communication System for Individuals with ALS
IEEE EMBC, 2025
[10]
Conference
TopoEEG: a TimeSformer-Based Topographic Image Representation Method for Single-Trial Early Detection of P300
IEEE ISBI, 2025
[11]
Workshop
ChatBCI: A Fast P300 Speller BCI Incorporating Generative AI-Based Word Prediction
IEEE Brain Discovery and Neurotechnology Workshop, 2024
[12]
Conference
P3T: A Transformer Model for Enhancing Character Recognition Rates in P300 Speller Systems
Asilomar ACSSC, 2024, pp. 514–518
[13]
Conference
IEEE EMBC, 2022, Glasgow, Scotland, pp. 3550–3553
Manuscripts currently under review.
[1]
Under Review
CORA: Coordinate-Aware ROI Attention for EEG Representation Learning in Cortical Space
Journal manuscript under review
[2]
Under Review
Subject-specific Channel Selection Based on Davies-Bouldin Index for EEG Motor Imagery Classification
Journal manuscript under review
[3]
Under Review
A Geometric Spatio-Learnable EEG Foundation Model Across Visual and Auditory Modalities in the Temporal Bisection Task
Journal manuscript under review
🎓 Advisor
🤝 Collaborators
Ashwini Subramanian
Ph.D. Student, ECE
Rutgers University
🧑🎓 Supervised Students
Pradyumna Rao
Ph.D. Student, ECE
Rutgers · 2025–
Logan Pasternak
Ph.D. Student, ECE
Rutgers · 2025–
Siwei Mai
Ph.D. Student, ECE
Rutgers · 2024–
Ankith Kunduru
Undergraduate Student, ECE
Rutgers · 2026–
Joel Paley
Master Student, ECE
Rutgers · 2024
Justin Ding
Master Student, ECE
Rutgers · 2023
William Milne
Master Student, ECE
Rutgers · 2023