Forest structure estimation is very important in geological, ecological and environmental studies. It provides the basis for the carbon stock estimation and effective means of sequestration of carbon sources and sinks. Multiple parameters are used to estimate the forest structure like above ground biomass, leaf area index and diameter at breast height. Among all these parameters, vegetation height has unique standing. In addition to forest structure estimation it provides the insight into long term historical changes and the estimates of stand age of the forests as well. There are multiple techniques available to estimate the canopy height. Light detection and ranging (LiDAR) based methods, being the accurate and useful ones, are very expensive to obtain and have no global coverage. There is a need to establish a mechanism to estimate the canopy height using freely available satellite imagery like Landsat images. Multiple studies are available which contribute in this area. The majority use Landsat images with random forest models. Although random forest based models are widely used in remote sensing applications, they lack the ability to utilize the spatial association of neighboring pixels in modeling process. In this research work, we define Convolutional Neural Network based model and analyze that model for three test configurations. We replicate the random forest based setup of Grant et al., which is a similar state-of-the-art study, and compare our results and show that the convolutional neural networks (CNN) based models not only capture the spatial association of neighboring pixels but also outperform the state-of-the-art.

Summary: With ever growing urbanization, the environmental noise is becoming hazardous. Vehicular traffic, locomotives, heavy machinery in industry, and construction processes are the major sources of noise pollution. It has adverse effects on the health of humans as well as that of the wild life. World Health Organization (WHO) puts noise pollution as the second major cause of illness due to environmental reasons. The effects of noise pollution on the quality of life are usually ignored. Due to this reason it is common, even in the first world countries, to have the WHO's peak noise standards violated in residential areas. Therefore, there is a need to have a real time, portable and easy to replicate, mechanism to monitor the noise sources. In this work, we propose a novel architecture of a deep neural network to classify a 10‐class environmental noise data called URBANSOUND8K. This network is comprised of three components, (1) one dimensional two level Discrete Wavelet Transform (DWT) component, (2) branched component for feature extraction through auto‐encoders, and (3) LSTM and fully‐connected layers based classification component. With all components combined, we call this network DWTNet. By embedding the DWT component as a part of network, we eliminate the need of prior data conversion into spectral and/or spectro‐temporal domains. The efficiency of DWTNet is comparable to the state of the art networks with significantly lower number of trainable parameters. We analyze the contribution of classification accuracy. We further study some of the classification results individually and show that some of the mis‐classifications are actually multi‐class classifications with distributed decision confidence.

Parkinson disease (PD) is a neurodegenerative disorder which is caused by the untimely death of dopamine generating neurons situated in substantia nigra pars compacta and striatum. The depletion of striatal dopamine induces the synchronization in brain activity in motor regions which results in tremors. Deep brain stimulation (DBS) is used as an obvious choice in therapy when the medication stops working. Traditional DBS systems are open loop systems in which stimulation parameters need to be adjusted manually from among a large pool of available combinations. This makes the whole process tedious for the doctors and traumatic for the patients. We have developed an FPGA based device which can monitor, learn and mimic the behavior of chaotic systems on the go. The purpose of this device is to learn the chaotic behavior of brain, when in normal state, and generate the chaotic stimulation signal during PD based tremors.

Deep brain stimulation (DBS) becomes the therapy of choice in the later stages of Parkinson's disease (PD) due to the medication's side effects. For effective DBS treatment, it is important to have a controlled dosage of DBS. DBS dosage is administered using the tuning of electrical parameters of the stimulus signal. Since this tuning process is tedious, time-consuming and patient-specific, there is a need to study the properties of DBS stimulation signal for proper dose administration. We propose a simulation framework to define an optimized DBS stimulus using Electroencephalogram (EEG) signals. The objective is to provide a simulation environment inspired by a realistic brain. The framework uses spiking neurons in a reservoir modelled after real brain anatomy and is trained using a biologically inspired spike-time-dependent-plasticity learning algorithm. This reservoir is initially set to OFF-medication state and forced to drift to the ON-medication state by optimizing the synaptic changes. In later testing, the generalization of this framework is verified with EEG-inverse solutions like sLORETA which utilize time-domain EEG signals to estimate neural activations. The stimulus signal is generated by accumulating the variations in synaptic weights in the neural reservoir in the target brain region. We analyze this signal and show that the application of this signal as stimulus results in decreased $\beta$ -band power in subthalamic-nucleus local-field-potential compared to OFF-medication local-field-potential without stimulation. Using SIM4LIFE simulation software, we show that the simulation increases chaos in the local-field-potential of subthalamic-nucleus neurons and shows that neuron weight variations follow specific trajectories in reconstructed state-space. Impact Statement: The primary contribution of this paper is to present a framework that utilizes the association between target signals of any modality (EEG, MEG, fMRI) and their respective post-synaptic activations to optimize the deep brain stimulation (DBS) stimulus signal. This framework combines coordinates of neural sources with patients' EEG data to optimize the stimulus of DBS. While defining this framework, we initially proposed the usage of spiking neurons in a biological neural reservoir to estimate the internal activations for given EEG data. Later, we extended our testing to incorporate inverse EEG algorithms that estimate neural activations directly from EEG data without any conversion. This is a novel framework, and there are not many examples available in the literature for similar work. This framework is in its early stages, yet it provides an adequate starting point for an under-explored area of study. This framework can potentially evolve into a useful tool for doctors in assisting the DBS tuning process before the surgical procedure. This framework's accuracy depends on the underlying technique to estimate the neural activations from the target signals (like EEG etc.).