Dr. Matthew Roos is a neuroscientist and engineer. He currently works with deep learning models for computer vision. His interests include advancing AI based on neuroscience, neuroevolution and artificial life, and brain-computer interfaces.
Prior to Binary Cognition, Dr. Roos worked at the Johns Hopkins University Applied Physics Lab. His work involved (1) utilizing neuromimetic AI models for processing of images, video, speech, etc., (2) developing brain-computer interfaces and exploring their applications, and (3) using neuroscience to predict and/or improve individual human performance on a range of tasks.
Dr. Roos obtained his PhD at the Johns Hopkins University Department of Neuroscience, where he studied the function of inhibitory neurotransmitters in the rodent auditory system.
- Johns Hopkins University, PhD, Neuroscience – Advisor: Dr. Bradford J. May
- Duke University, MS, Biomedical Engineering – Advisor: Dr. Edward W. Hsu
- Purdue University, MS, Electrical and Computer Engineering – Advisor: Dr. Michael P. Fitz
- University of Missouri-Rolla, BS, Electrical Engineering – Advisor: Dr. Kurt Kosbar
Reilly, et al. (2017) describes and justifies the Neural Reconstruction Integrity (NRI) metric for assessing the connectivity of reconstructed neural networks. Code that computes the metric and applies it to a toy example is found here.
Reilly EP, Garretson JS, Gray Roncal W, Kleissas DM, Wester BA, Chevillet MA, Roos MJ (2018) Neural Reconstruction Integrity: A metric for assessing the connectivity accuracy of reconstructed neural networks. Front Neuroinform 12:74.
Ratto C, Caceres C, Roos M, Rupp K, Milsap G, Crone N, Wolmetz M (2017) Approaches to zero-shot stimulus decoding in ECoG for potential BCI applications. Proc of Graz BCI Conference 2017.
Caceres CA, Roos MJ, Rupp KM, Milsap G, Crone NE, Wolmetz ME, Ratto, CR (2017) Feature Selection Methods for Zero-Shot Learning of Neural Activity. Front Neuroinform 11:41.
Rupp K, Roos M, Milsap G, Caceres C, Ratto C, Chevillet M, Crone NE, Wolmetz M (2017) Semantic attributes are encoded in human electrocorticographic signals during visual object recognition. NeuroImage. 148: 318-329.
Glasgow K, Roos M, Haufler A, Chevillet M, Wolmetz M (2016) Evaluating semantic models with word-sentence relatedness. arXiv preprint arXiv:1603.07253.
Roos MJ (2013) Spectral and temporal coding in the ventral cochlear nucleus and the role of inhibition. The Johns Hopkins University, PhD thesis.
Roos MJ, May BJ (2012) Classification of unit types in the anteroventral cochlear nucleus of laboratory mice. Hear Res. 289(1-2): 13-26.
May BJ, Lauer AM, Roos MJ (2011) Impairments of the medial olivocochlear system increase the risk of noise-induced auditory neuropathy in laboratory mice. Otol Neurotol. 32(9): 1568-78.
May BJ, Anderson M, Roos M (2008) The role of broadband inhibition in the rate representation of spectral cues for sound localization in the inferior colliculus. Hear Res. 238(1-2): 77-93.
Roos MJ (2002) A study of acquisition time efficiency of magnetic resonance diffusion tensor imaging. Duke University, MS thesis.
Fitz MP, Roos MJ, Grimm J, Krogmeier JV (1997) A high performance point-to-point modem for the ITS 220 MHz spectral allocation. Intelligent Transportation System, 1997. ITSC ’97., IEEE Conference on: 572-7
Fitz MP, Krogmeier JV, Grimm J, Chen T-A, Gansman JA, Roos M, Emsley B, Blackburn D, Muh C (1997) ITS wireless narrowband land mobile data communications. Vehicular Technology Conference, 1997, IEEE 47th (2): 909-13.
Roos MJ (1997) Architectural design and implementation of a wireless narrowband fixed-point modem. Purdue University, MS thesis.
Milsap GW, Collard MJ, Rupp K, Roos MJ, Caceres C, Ratto C, Wolmetz M, Crone NE (2017) Intrinsic neural spaces from human electrocorticography. Soc Neurosci Abst.
Roos MJ, Reilly EP, Garretson JS, Gray Roncal W, Kleissas DM, Chevillet MA, Wester BA (2016) Neural Reconstruction Integrity: A novel connectomics metric sensitive to brain graph connectivity rather than fine image segmentation. Soc Neurosci Abst.
Roos MJ, Wolmetz M, Chevillet MA (2014) A hierarchical model of vision (HMAX) can also recognize speech. Org Comp Neuro Abst.
Roos M, May B (2012) Effects of glycinergic inhibition on spectral and temporal coding in the rat’s ventral cochlear nucleus. Assoc Res Oto Abst.
Roos M, May B (2011) Physiological classification of single-unit activity in ventral cochlear nucleus of the mouse. Assoc Res Oto Abst.
Roos M, Lauer L, May B (2009) Temporal coding in the ventral cochlear nucleus of mice with impaired cochlear efferent transmission. Soc Neurosci Abst.
Roos M, May B, Doucet J (2009) The innervation of T-stellate multipolar neurons in ventral cochlear nucleus. Assoc Res Oto Abst.
Grimm J, Krogmeier JV, Chen T-A, Roos M, Emsley B, Blackburn D, Fitz MP, Gansman J (1997) Field testing on the ITS 220 MHz narrowband channels. ITS America Seventh Annual Meeting and Exposition Abst.
“Neurons, Brains, and Neuromimetic Machine Learning.” Baltimore AI, Baltimore, MD. January 3, 2019. Announcement here.
“Research program in applied neuroscience: Motivation, goals, and progress.” The Laboratory for Physical Sciences, College Park, MD. May 17, 2017.
“Effects of glycinergic inhibition on spectral and temporal coding in the rat’s ventral cochlear nucleus.” Center for Hearing and Balance, The Johns Hopkins University, Baltimore, MD. April 5, 2012.
“Spectral and temporal coding in the ventral cochlear nucleus: The role of inhibition.” The Johns Hopkins Applied Physics Laboratory, Baltimore, MD. September 10, 2012.