A5052682430- Neural dynamics and brain function
- Photorefractive and Nonlinear Optics
- Neural Networks and Applications
- Gender Studies in Language
- Face Recognition and Perception
- Solid State Laser Technologies
- Memory Processes and Influences
- Advanced Fiber Laser Technologies
- Luminescence Properties of Advanced Materials
- Functional Brain Connectivity Studies
- Photonic and Optical Devices
- Memory and Neural Mechanisms
- Anomaly Detection Techniques and Applications
- Crystal Structures and Properties
- Topic Modeling
- Solid-state spectroscopy and crystallography
- Face and Expression Recognition
- Advanced Neuroimaging Techniques and Applications
- Advanced Semiconductor Detectors and Materials
- Visual perception and processing mechanisms
- Advanced Neural Network Applications
- Authorship Attribution and Profiling
- Optical and Acousto-Optic Technologies
- Machine Learning in Healthcare
- Multimodal Machine Learning Applications
Princeton University
2018-2024
Columbia University
2024
Neuroscience Institute
2020-2022
Shandong University
2001-2021
Recent human behavioral and neuroimaging results suggest that people are selective in when they encode retrieve episodic memories. To explain these findings, we trained a memory-augmented neural network to use its memory support prediction of upcoming states an environment where past situations sometimes reoccur. We found the learned selectively as function several factors, including uncertainty about state. Additionally, encoding memories at end event (but not mid-event) led better...
Advanced brain imaging analysis methods, including multivariate pattern (MVPA), functional connectivity, and alignment, have become powerful tools in cognitive neuroscience over the past decade. These are implemented custom code separate packages, often requiring different software language proficiencies. Although usable by expert researchers, novice users face a steep learning curve. difficulties stem from use of new programming languages (e.g., Python), how to apply machine-learning...
How does the human brain encode semantic information about objects? This paper reconciles two seemingly contradictory views. The first proposes that local neural populations independently features; second, representations arise as a dynamic distributed code changes radically with stimulus processing. Combining simulations well-known network model of memory, multivariate pattern classification, and electrocorticography, we find both views are partially correct: animacy depicted is across...
Functional magnetic resonance imaging (fMRI) offers a rich source of data for studying the neural basis cognition. Here, we describe Brain Imaging Analysis Kit (BrainIAK), an open-source, free Python package that provides computationally optimized solutions to key problems in advanced fMRI analysis. A variety techniques are presently included BrainIAK: intersubject correlation (ISC) and functional connectivity (ISFC), alignment via shared response model (SRM), full matrix analysis (FCMA),...
Functional magnetic resonance imaging (fMRI) offers a rich source of data for studying the neural basis cognition. Here, we describe Brain Imaging Analysis Kit (BrainIAK), an open-source, free Python package that provides computationally-optimized solutions to key problems in advanced fMRI analysis. A variety techniques are presently included BrainIAK: intersubject correlation (ISC) and functional connectivity (ISFC), alignment via shared response model (SRM), full matrix analysis (FCMA),...
Abstract It has been proposed that, when processing a stream of events, humans divide their experiences in terms inferred latent causes (LCs) to support context-dependent learning. However, shared structure is present across contexts, it still unclear how the “splitting” LCs and learning can be simultaneously achieved. Here, we Latent Cause Network (LCNet), neural network model LC inference. Through learning, naturally stores that tasks weights. Additionally, represents context-specific...
Abstract Generalization to new tasks requires learning of task representations that accurately reflect the similarity structure space. Here, we argue episodic memory (EM) plays an essential role in this process by stabilizing representations, thereby supporting accumulation structured knowledge. We demonstrate using a neural network model infers minimize current task’s objective function; crucially, can retrieve previously encoded from EM and use these initialize inference process. With EM,...
Abstract Recent human behavioral and neuroimaging results suggest that people are selective in when they encode retrieve episodic memories. To explain these findings, we trained a memory-augmented neural network to use its memory support prediction of upcoming states an environment where past situations sometimes reoccur. We found the learned selectively as function several factors, including uncertainty about state. Additionally, encoding memories at end event (but not mid-event) led better...
Through specific experiences, humans learn the relationships that underlie structure of events in world. Schema theory suggests we organize this information mental frameworks called “schemata,” which represent our knowledge Generalizing structural to new situations requires role-filler binding, ability associate “fillers” with abstract “roles.” For instance, when hear sentence Alice ordered a tea from Bob , bindings customer:Alice drink:tea and barista:Bob allow us understand make inferences...
Different neural networks trained on the same dataset often learn similar input-output mappings with very different weights. Is there some correspondence between these network solutions? For linear networks, it has been shown that instances of architecture encode representational similarity matrix, and their activity patterns are connected by orthogonal transformations. However, is unclear if this holds for non-linear networks. Using a shared response model, we show input examples as...
Rare-earths (Pr, Ho) doped double tungstates NaY(WO4)2 crystals were prepared by using Czochralski (CZ) pulling method. Their absorption spectra have been measured in the region 300-2000nm at room temperature. A band gap of eigentransition 3.92eV was obtained. The three phenomenological parameters calculated through fitting and oscillator strengths. J-O theory used to evaluate spontaneous emission probabilities, stimulated cross sections, branch ratios energy level lifetimes some useful...
Abstract The paper reports on the degradation of InGaN/GaN Blue LED submitted to proton irradiation at 80 MeV and various fluences (4×10 13 p cm −2 1×10 14 ). After irradiation, we found a decrease in light output power external quantum efficiency with fluence. Photoluminescence (PL) measurements exhibited that peak position 400, 447 568 nm remained unchanged, only intensity decreased. blue emission reduced by 75%, indicating active region degraded seriously; from In x Ga 1- N ( = 0.11) more...
Double waveguides were produced in potassium titanyl phosphate (KTiOPO4 or KTP) by ion exchange pure RbNO3 at 340 °C for 45 min, and subsequently implantation with 500 keV O ions a dose of 1 × 1015 cm−2. By the prism-coupling method guided mode spectra as well refractive index profiles ion-exchanged sample determined. The relative value distribution implanted KTP z direction was simulated, which indicates formation double waveguides. field pattern propagation light waveguide collected...
The 1064 nm optimal spatial phase matching (PM) direction for LaCa4O(BO3)3 (LaCOB) crystals was found to be (113.5°, 43.8°) with deff of 1.34 pm V−1, angular acceptance 0.9 mrad cm and walk-off angle 14.9 mrad, more than a 56% second-harmonic generation (SHG) conversion efficiency realized on sample (4 × 4 6 mm3) cut in this direction, by using Nd : YAG pico-second laser. noncritical phase-matching (A-NCPM) wavelengths along the y z axes LaxGd1−xCOB (x = 0.09, 0.13) were measured. Using...
Abstract Li 2 O·3B O 3 or O(B ) (LOBO) crystallizes in the orthorhombic space group Pna 1 with cell parameters a = 0.73788(6), b 0.84473(7) and c 0.51395(5) nm two formula units primitive cell. Raman results show that characteristic spectra of LOBO are mainly contributed by particularly B(3)—O tetrahedra also B(1)—O B(2)—O triangles partly ascribed to breathing vibrations B—O rings. The structural rigidity is associated all bond stretching bending especially bonds. excellent non‐linear...
Abstract How does the human brain encode semantic information about objects? This paper reconciles two seemingly contradictory views. The first proposes that local neural populations independently features; second, representations arise as a dynamic distributed code changes radically with stimulus processing. Combining simulations well-known network model of memory, multivariate pattern classification, and electrocorticography, we find both views are partially correct: is across ventral...