Recent work indicates that low-dimensional dynamics of neural and behavioral data are often preserved across days and subjects. However, extracting these preserved dynamics remains challenging: high-dimensional neural population activity and the recorded neuron populations vary across recording sessions. While existing modeling tools can improve alignment between neural and behavioral data, they often operate on a per-subject basis or discretize behavior into categories, disrupting its natural continuity and failing to capture the underlying dynamics. We introduce $\underline{\text{C}}$ontrastive $\underline{\text{A}}$ligned $\underline{\text{N}}$eural $\underline{\text{D}}$$\underline{\text{Y}}$namics (CANDY), an end‑to‑end framework that aligns neural and behavioral data using rank-based contrastive learning, adapted for continuous behavioral variables, to project neural activity from different sessions onto a shared low-dimensional embedding space. CANDY fits a shared linear dynamical system to the aligned embeddings, enabling an interpretable model of the conserved temporal structure in the latent space. We validate CANDY on synthetic and real-world datasets spanning multiple species, behaviors, and recording modalities. Our results show that CANDY is able to learn aligned latent embeddings and preserved dynamics across neural recording sessions and subjects, and it achieves improved cross-session behavior decoding performance. We further show that the latent linear dynamical system generalizes to new sessions and subjects, achieving comparable or even superior behavior decoding performance to models trained from scratch. These advances enable robust cross‑session behavioral decoding and offer a path towards identifying shared neural dynamics that underlie behavior across individuals and recording conditions. The code and two-photon imaging data of striatal neural activity that we acquired here are available at https://github.com/schnitzer-lab/CANDY-public.git.

Abstract The novel network contains many sensors, which greatly heightens data transmission burdens. Some networks require the data perceived by sensors for a period to make decisions. Drawing inspiration from the human neural conduction mechanism, a waveform data encoding method called feature sensing neural coding (FSNC) is proposed to enhance network data transmission efficiency. It involves feature decomposition of information and subsequent non‐linear encoding of feature coefficients for data transmission. This approach exploits the unique neuronal responses to diverse stimuli and the inherent non‐linear characteristics of human neural coding. Finally, taking the speech signal and seismic wave signal as examples, the effectiveness of FSNC is verified by simulating the auditory nerve conduction process with frequency as a feature according to the mechanism of travelling wave motion of the basilar membrane in the cochlea. Moreover, experiments on seismic waveform signals have demonstrated the wide applicability of FSNC. Compared with traditional speech coding schemes, the FSNC bit rate is only 6.4 kbps, which greatly reduces the amount of data transmitted. Not only that, FSNC also has a certain fault tolerance, and parallel transmission can also greatly increase the transmission rate. This research provides new ideas for efficient data transmission over new networks.

Abstract There are a large number of perceptrons in the new bionic network. To improve the efficiency of data transmission in the bionic network, a maximum entropy neural coding method is proposed. By drawing on the characteristics of human nerve conduction, the authors designed a data transmission model and adopted an adaptive spike firing rate encoding strategy to maximize information entropy, thereby improving encoding efficiency. The simulation experiment results and the applications of the maximum entropy neural coding method to fault detection and seismic detection have validated the effectiveness of the maximum entropy neural coding method. Even if there is certain data distortion, the statistical characteristics of the decoded data and the fault detection performance will not be affected. This research not only proposes novel approaches for efficient data transmission in bionic network, but also identifies possible directions for enhancing data transmission efficiency through the integration of task‐oriented semantic communications in future applications.

Tensor computation plays a paramount role in a broad range of domains, including machine learning, data analytics, and scientific computing. The wide adoption of tensor computation and its huge computation cost has led to high demand for flexible, portable, and high-performance library implementation on heterogeneous hardware accelerators such as GPUs and FPGAs. However, the current tensor library implementation mainly requires programmers to manually design low-level implementation and optimize from the algorithm, architecture, and compilation perspectives. Such a manual development process often takes months or even years, which falls far behind the rapid evolution of the application algorithms.