Research
Google Scholar: Xin-Ya Zhang
Xin-Ya Zhang, S. Bobadilla-Suarez, X. Luo, M. Lemonari, S. L. Brincat, M. Siegel, E. K. Miller and B. C. Love. bioRxiv.
We found that except for V4 (color bound) and MT (motion bound), the brain radically reconfigured itself to stretch representations along task-relevant dimensions in multiple regions. A deep learning model (CNN-LSTM) was trained on the same visual input and rewards as the monkeys. Despite lacking an explicit selective attention or other control mechanism, the model displayed task-relevant stretching as a consequence of error minimization, indicating that stretching is an adaptive strategy.
We found that except for V4 (color bound) and MT (motion bound), the brain radically reconfigured itself to stretch representations along task-relevant dimensions in multiple regions. A deep learning model (CNN-LSTM) was trained on the same visual input and rewards as the monkeys. Despite lacking an explicit selective attention or other control mechanism, the model displayed task-relevant stretching as a consequence of error minimization, indicating that stretching is an adaptive strategy.
Xin-Ya Zhang, H. Lin, Z. Deng, M. Siegel, E. K. Miller, G. Yan. bioRxiv.
When we encounter familiar environments, the brain may trigger neuronal activities that resemble specific patterns. Our findings show that the machine effectively captures the function of individual brain regions as a consequence of minimizing visual errors. Specifically, in areas related to vision, V4 and LIP are implicated in visual color and shape processing, while MT is more involved in visual motion. Moreover, we highlight the importance of the source brain region in both decoding and encoding processes. While each brain region harbors some global information, achieving good performance in vision reconstruction necessitates utilizing data from regions closely linked with vision processing.
When we encounter familiar environments, the brain may trigger neuronal activities that resemble specific patterns. Our findings show that the machine effectively captures the function of individual brain regions as a consequence of minimizing visual errors. Specifically, in areas related to vision, V4 and LIP are implicated in visual color and shape processing, while MT is more involved in visual motion. Moreover, we highlight the importance of the source brain region in both decoding and encoding processes. While each brain region harbors some global information, achieving good performance in vision reconstruction necessitates utilizing data from regions closely linked with vision processing.
3.Geometric scaling law in real neuronal networks
Xin-Ya Zhang, J. M. Moore, X. Ru and G. Yan. submitted.
We uncovered a fundamental and elegant scaling law in the synapse-resolution Drosophila connectomes. This discovery challenges the well-known exponential distance rule previously established in inter-areal brain networks and carries functional significance, aligning with the maximum entropy of information communication and the functional criticality balancing integration and segregation. Our findings establish a direct link between brain geometry and topology, hinting at new opportunities for developing brain geometry-inspired artificial intelligence.
We uncovered a fundamental and elegant scaling law in the synapse-resolution Drosophila connectomes. This discovery challenges the well-known exponential distance rule previously established in inter-areal brain networks and carries functional significance, aligning with the maximum entropy of information communication and the functional criticality balancing integration and segregation. Our findings establish a direct link between brain geometry and topology, hinting at new opportunities for developing brain geometry-inspired artificial intelligence.
X. Ru, X.-Y. Zhang, Zijia Liu, J. M. Moore and G. Yan. NeurIPS 2023.
We consider the problem of reconstructing coupled networks (e.g., biological neural networks) connecting large numbers of variables (e.g.,nerve cells), of which state evolution is governed by dissipative dynamics consisting of strong self-drive (dominants the evolution) and weak coupling-drive. The core difficulty is sparseness of coupling effect that emerges (the coupling force is significant) only momentarily and otherwise remains quiescent in time series (e.g., neuronal activity sequence). Here we learn the idea from attention mechanism to guide the classifier to make inference focusing on the critical regions of time series data where coupling effect may manifest. Specifically, attention coefficients are assigned autonomously by artificial neural networks trained to maximise the Attentive Transfer Entropy (ATEn), which is a novel generalization of the iconic transfer entropy metric. Our results show that, without any prior knowledge of dynamics, ATEn explicitly identifies areas where the strength of coupling-drive is distinctly greater than zero. This innovation substantially improves reconstruction performance for both synthetic and real directed coupling networks using data generated by neuronal models widely used in neuroscience.
We consider the problem of reconstructing coupled networks (e.g., biological neural networks) connecting large numbers of variables (e.g.,nerve cells), of which state evolution is governed by dissipative dynamics consisting of strong self-drive (dominants the evolution) and weak coupling-drive. The core difficulty is sparseness of coupling effect that emerges (the coupling force is significant) only momentarily and otherwise remains quiescent in time series (e.g., neuronal activity sequence). Here we learn the idea from attention mechanism to guide the classifier to make inference focusing on the critical regions of time series data where coupling effect may manifest. Specifically, attention coefficients are assigned autonomously by artificial neural networks trained to maximise the Attentive Transfer Entropy (ATEn), which is a novel generalization of the iconic transfer entropy metric. Our results show that, without any prior knowledge of dynamics, ATEn explicitly identifies areas where the strength of coupling-drive is distinctly greater than zero. This innovation substantially improves reconstruction performance for both synthetic and real directed coupling networks using data generated by neuronal models widely used in neuroscience.
X. Ru, J. M. Moore, X.-Y. Zhang, Y. Zeng and G. Yan. AAAI 2023.
The world is currently seeing frequent local outbreaks of epidemics, such as COVID-19 and Monkeypox. Preventing further propagation of the outbreak requires prompt implementation of control measures, and a critical step is to quickly infer patient zero. To address these challenges, we tailor a GNN-based model to establish the inverse statistical association between the current and initial state implicitly. We also demonstrate that our method is robust to missing information about contact structure or current state.
The world is currently seeing frequent local outbreaks of epidemics, such as COVID-19 and Monkeypox. Preventing further propagation of the outbreak requires prompt implementation of control measures, and a critical step is to quickly infer patient zero. To address these challenges, we tailor a GNN-based model to establish the inverse statistical association between the current and initial state implicitly. We also demonstrate that our method is robust to missing information about contact structure or current state.
X.-Y. Zhang, J. Sun and G. Yan. Physical Review Research, 3(3), p.L032045. 2021.
We explored a general model of temporal networks and analytically proved that the weight variation of a link is equivalent to attaching a virtual driver node to that link. Consequently, the temporality of link weights can significantly increase the dimension of controllable space and remarkably reduce control cost.
We explored a general model of temporal networks and analytically proved that the weight variation of a link is equivalent to attaching a virtual driver node to that link. Consequently, the temporality of link weights can significantly increase the dimension of controllable space and remarkably reduce control cost.
Research experience
Geometric scaling law in real neuronal networks
Oral presentation at 20th China Networks Science Forum 2024.
Attentive Transfer Entropy to Exploit Transient Emergence of Coupling Effect
Spotlight at International Conference NeurIPS 2023.
Link weight variation offers superiority in controlling temporal networks
Oral presentation at International Conference NetSci 2022.
Structural origin of co-susceptibility in cascading failures
Oral presentation at International Conference NetSci-X 2018. We found that both structural closeness and high-order correlations could lead to co-susceptibility. This finding prompted us to propose a new statistical quantity, based on structure only, to assess the co-susceptibility of node pairs in an arbitrary network.
Oral presentation at International Conference NetSci-X 2018. We found that both structural closeness and high-order correlations could lead to co-susceptibility. This finding prompted us to propose a new statistical quantity, based on structure only, to assess the co-susceptibility of node pairs in an arbitrary network.
Working Papers
• Data-driven control of nonlinear network systems
While most previous work of network control focused on linearized systems that represent local dynamics around a basin, real network systems exhibit multiple basins or attractors. Hence it is of importance to find the strategy to steer the network system from one basin to any desired basin. In this project we are developing a deep learning framework for controlling nonlinear systems. The method is model-free, i.e. based on the observation of nodal activities only.
While most previous work of network control focused on linearized systems that represent local dynamics around a basin, real network systems exhibit multiple basins or attractors. Hence it is of importance to find the strategy to steer the network system from one basin to any desired basin. In this project we are developing a deep learning framework for controlling nonlinear systems. The method is model-free, i.e. based on the observation of nodal activities only.