Optimal control methods for nonlinear parameter estimation in biophysical neuron models

Functional forms of biophysically-realistic neuron models are constrained by neurobiological and anatomical considerations, such as cell morphologies and the presence of known ion channels. Despite these constraints, neurons models still contain unknown static parameters which must be inferred from experiment. This inference task is most readily cast into the framework of state-space models, which systematically takes into account partial observability and measurement noise. Inferring only dynamical state variables such as membrane voltages is a well-studied problem, and has been approached with a wide range of techniques beginning with the well-known Kalman filter. Inferring both states and fixed parameters, on the other hand, is less straightforward. Here, we develop a method for joint parameter and state inference that combines traditional state space modeling with chaotic synchronization and optimal control. Our methods are tailored particularly to situations with considerable measurement noise, sparse observability, very nonlinear or chaotic dynamics, and highly uninformed priors. We illustrate our approach both in a canonical chaotic model and in a phenomenological neuron model, showing that many unknown parameters can be uncovered reliably and accurately from short and noisy observed time traces. Our method holds promise for estimation in larger-scale systems, given ongoing improvements in calcium reporters and genetically-encoded voltage indicators.

Inter-Nesting Movements, Migratory Pathways, and Resident Foraging Areas of Green Sea Turtles (Chelonia mydas) Satellite-Tagged in Southwest Florida

Globally, sea turtle research and conservation efforts are underway to identify important high-use areas where these imperiled individuals may be resident for weeks to months to years. In the southeastern Gulf of Mexico, recent telemetry studies highlighted post-nesting foraging sites for federally endangered green turtles (Chelonia mydas) around the Florida Keys. In order to delineate additional areas that may serve as inter-nesting, migratory, and foraging hotspots for reproductively active females nesting in peninsular southwest Florida, we satellite-tagged 14 green turtles that nested at two sites along the southeast Gulf of Mexico coastline between 2017 and 2019: Sanibel and Keewaydin Islands. Prior to this study, green turtles nesting in southwest Florida had not previously been tracked and their movements were unknown. We used switching state space modeling to show that an area off Cape Sable (Everglades), Florida Bay, and the Marquesas Keys are important foraging areas that support individuals that nest on southwest Florida mainland beaches. Turtles were tracked for 39–383 days, migrated for a mean of 4 days, and arrived at their respective foraging grounds in the months of July through September. Turtles remained resident in their respective foraging sites until tags failed, typically after several months, where they established mean home ranges (50% kernel density estimate) of 296 km2. Centroid locations for turtles at common foraging sites were 1.2–36.5 km apart. The area off southwest Florida Everglades appears to be a hotspot for these turtles during both inter-nesting and foraging; this location was also used by turtles that were previously satellite tagged in the Dry Tortugas after nesting. Further evaluation of this important habitat is warranted. Understanding where and when imperiled yet recovering green turtles forage and remain resident is key information for designing surveys of foraging resources and developing additional protection strategies intended to enhance population recovery trajectories.

3D City Online Visualization and Cluster Architecture for Digital City

As an important carrier of human production, life, and social development, the emergence of cities symbolizes the maturity and civilization of mankind. For more than 40 years of reform and opening up, our country’s economic development has become increasingly prosperous, and urbanization is booming. At present, our country is in a decisive period for building a well-off society in an all-round way, with rapid progress in socio-economic growth and urbanization. Based on this, this article is oriented towards urban visualization modeling work and proposes a cluster modeling method that is compatible with the combination of urban geological structure and three-dimensional urban space, so that urban space modeling work not only pays attention to the rationality of above-ground planning and construction but also fully considers underground geology the stability and safety of the structure. This paper uses the 3D city online visualization modeling technology to efficiently and reasonably complete the 3D urban geological modeling under the fusion of multiple geological data and combines the organic combination of multisource heterogeneous model data to convert the geological model data into a 3D geographic information model; the universal standard format analyzes the rapid construction of large-scale complex geological structure models and the integrated expression of multisource heterogeneous model data. Experiments have proved that from the cluster capacity of 5,000 to 100,000, no matter how much the modeling time is different, whether it is to search the entire territory or part of the scope, the search time of the 3D city visualization model is less than 20 ms, and the 3D city visualization model map of the city can be well established. This shows that the three-dimensional city visualization model highlights the impact of the urban geological environment on urban construction and development and visually and vividly displays region geological structure and other information in a three-dimensional way, providing corresponding information for urban geological stability assessment and geological disaster rescue reference and help.

Dissimilarity space reinforced with manifold learning and latent space modeling for improved pattern classification

Dissimilarity representation plays a very important role in pattern recognition due to its ability to capture structural and relational information between samples. Dissimilarity space embedding is an approach in which each sample is represented as a vector based on its dissimilarity to some other samples called prototypes. However, lack of neighborhood-preserving, fixed and usually considerable prototype set for all training samples cause low classification accuracy and high computational complexity. To address these challenges, our proposed method creates dissimilarity space considering the neighbors of each data point on the manifold. For this purpose, Locally Linear Embedding (LLE) is used as an unsupervised manifold learning algorithm. The only goal of this step is to learn the global structure and the neighborhood of data on the manifold and mapping or dimension reduction is not performed. In order to create the dissimilarity space, each sample is compared only with its prototype set including its k-nearest neighbors on the manifold using the geodesic distance metric. Geodesic distance metric is used for the structure preserving and is computed using the weighted LLE neighborhood graph. Finally, Latent Space Model (LSM), is applied to reduce the dimensions of the Euclidean latent space so that the second challenge is resolved. To evaluate the resulted representation ad so called dissimilarity space, two common classifiers namely K Nearest Neighbor (KNN) and Support Vector Machine (SVM) are applied. Experiments on different datasets which included both Euclidean and non-Euclidean spaces, demonstrate that using the proposed approach, classifiers outperform the other basic dissimilarity spaces in both accuracy and runtime.