A5009612938- Neural dynamics and brain function
- Advanced Memory and Neural Computing
- Neuroscience and Music Perception
- Neural Networks and Applications
- Cell Image Analysis Techniques
- Visual perception and processing mechanisms
- Advanced Vision and Imaging
- Neuroscience and Neuropharmacology Research
- Photoreceptor and optogenetics research
- Anomaly Detection Techniques and Applications
- Error Correcting Code Techniques
- Fractal and DNA sequence analysis
- Aesthetic Perception and Analysis
- AI in cancer detection
- Memory and Neural Mechanisms
- Cancer Genomics and Diagnostics
- Mathematical and Theoretical Epidemiology and Ecology Models
- Music and Audio Processing
- Cancer-related molecular mechanisms research
- Evolutionary Algorithms and Applications
- Robotics and Automated Systems
- Algorithms and Data Compression
- Blind Source Separation Techniques
- EEG and Brain-Computer Interfaces
- Evolutionary Game Theory and Cooperation
Qingdao University
2020-2024
Hong Kong Baptist University
2016-2022
Shanghai Center for Brain Science and Brain-Inspired Technology
2021
Institute for the Future
2021
Institute of Theoretical Physics
2011-2016
Chinese Academy of Sciences
2011-2016
In an iterated non-cooperative game, if all the players act to maximize their individual accumulated payoff, system as a whole usually converges Nash equilibrium that poorly benefits any player. Here we show such undesirable destiny is avoidable in Rock-Paper-Scissors (RPS) game involving two rational players, X and Y. Player has option of proactively adopting cooperation-trap strategy, which enforces complete cooperation from player Y leads highly beneficial maximally fair situation both...
To maximize future rewards in this ever-changing world, animals must be able to discover the temporal structure of stimuli and then anticipate or act correctly at right time. How perceive, maintain, use time intervals ranging from hundreds milliseconds multi-seconds working memory? information is processed concurrently with spatial decision making? Why there are strong neuronal signals tasks which not required? A systematic understanding underlying neural mechanisms still lacking. Here, we...
Training biophysical neuron models provides insights into brain circuits' organization and problem-solving capabilities. Traditional training methods like backpropagation face challenges with complex due to instability gradient issues. We explore evolutionary algorithms (EAs) combined heterosynaptic plasticity as a gradient-free alternative. Our EA agents distinct information routes, evaluated via alternating gating, guided by dopamine-driven plasticity. This model draws inspiration from...
Disordered and frustrated graphical systems are ubiquitous in physics, biology, information science. For models on complete graphs or random graphs, deep understanding has been achieved through the mean-field replica cavity methods. But finite-dimensional 'real' remain very challenging because of abundance short loops strong local correlations. A statistical mechanics theory is constructed this paper for based mathematical framework partition function expansion concept region graphs....
In neural systems, synaptic plasticity is usually driven by spike trains. Due to the inherent noises of neurons and synapses as well randomness connection details, trains typically exhibit variability such spatial temporal stochasticity, resulting in changes under plasticity, which we call efficacy variability. How influences remains unclear. this paper, try understand influence pair-wise additive spike-timing dependent (STDP) when mean strength plastic into a neuron bounded (synaptic...
Limited precision of synaptic weights is a key aspect both biological and hardware implementation neural networks. To assign low-precise during learning non-trivial task, but may benefit from representing to-be-learned items using sparse code. However, the computational difficulty resulting low weight advantage coding remain not fully understood. Here, we study perceptron model, which associates binary (0 or 1) input patterns with desired outputs weights, modeling single neuron receiving...
Synapses may undergo variable changes during plasticity because of the variability spike patterns such as temporal stochasticity and spatial randomness. Here, we call synaptic weight to be efficacy variability. In this paper, investigate how four aspects pattern statistics (i.e., synchronous firing, burstiness/regularity, heterogeneity rates cross-correlations) influence under pair-wise additive spike-timing dependent (STDP) homeostasis (the mean strength plastic synapses into a neuron is...
Cellular deconvolution via bulk RNA sequencing (RNA-seq) presents a cost-effective and efficient alternative to experimental methods such as flow cytometry single-cell RNA-seq (scRNA-seq) for analyzing the complex cellular composition of tumor microenvironments. Despite challenges due heterogeneity within among tumors, our innovative deep learning–based approach, DeSide, shows exceptional accuracy in estimating proportions 16 distinct cell types subtypes solid tumors. DeSide integrates...
Abstract Cellular decomposition employing bulk RNA-sequencing (RNA-seq) has been consistently under investigation due to its high fidelity, ease of use, and cost-effectiveness compared single cell (scRNA-seq). However, the intricate nature tumor microenvironment, significant heterogeneity among patients cells have made it challenging precisely evaluate cellular composition solid tumors using a unified model. In this work, we developed DeSide, deep learning single-cell method for tumors,...
Summary How the brain modifies synapses to improve performance of complicated networks remains one biggest mysteries in neuroscience. Canonical models suppose synaptic weights change according pre- and post-synaptic activities (i.e., local plasticity rules), implementing gradient-descent algorithms. However, lack experimental evidence confirm these suggests that there may be important ingredients neglected by models. For example, heterosynaptic plasticity, non-local rules mediated...
Abstract Perception or imagination requires top-down signals from high-level cortex to primary visual (V1) reconstruct simulate the representations bottom-up stimulated by seen images. Interestingly, in V1 have lower spatial resolution than representations. It is unclear why brain uses low-resolution high-resolution By modeling pathway of system using decoder variational auto-encoder (VAE), we reveal that can better information contained sparse activities simple cells, which facilitates...
In neural systems, synaptic plasticity is usually driven by spike trains. Due to the inherent noises of neurons, synapses and networks, trains typically exhibit externally uncontrollable variability such as spatial heterogeneity temporal stochasticity, resulting in synapses, which we call efficacy variability. Spike patterns with same population rate but inducing different may result neuronal networks sharply structures functions. However, how influences remains unclear. Here, systematically...
A pervasive research protocol of cognitive neuroscience is to train subjects perform deliberately designed experiments and record brain activity simultaneously, aiming understand the mechanism underlying cognition. However, how results this can be applied in technology seldom discussed. Here, I review studies on time processing as examples protocol, well two main application areas (neuroengineering brain-inspired artificial intelligence). Time an indispensable dimension cognition; also any...
Despite the importance of timing, our understanding is limited regarding how second-scale time perception mediated in human brain. Here we combined intracranial stereoelectroencephalography (SEEG) recordings subjects with circuit dissection mice to show that visual cortex encodes timing information. We first performed an interval task identify (V1) as a key brain area. then conducted optogenetic experiments necessary for behavior. Consistent this possibility, V1 neurons fire time-keeping...