We explore whether Neural Networks (NNs) can {\it discover} the presence of symmetries as they learn to perform a task. For this, we train hundreds NNs on decoy task} based well-controlled Physics templates, where no information symmetry is provided. use output from last hidden layer all these NNs, projected fewer dimensions, input for classification task, and show that had indeed been identifi...

We investigate questions of how to reason about learner meaning in cases where the set of correct meanings is never entirely complete, specifically for the case of picture description tasks (PDTs). To operationalize this, we explore different models of representing and scoring non-native speaker (NNS) responses to a picture, including bags of dependencies, automatically determining the relevant...

We improve previous end-to-end differentiable neural networks (NNs) with fast weight memories. A gate mechanism updates fast weights at every time step of a sequence through two separate outer-product-based matrices generated by slow parts of the net. The system is trained on a complex sequence to sequence variation of the Associative Retrieval Problem with roughly 70 times more temporal memory...

Observations of ozone from satellite platform is a key source of information to monitor and understand processes that may have a relevant impact on public health, air quality and climate. To achieve at the same time a reasonable spatial resolutions and a sufficient sensitivity, nadir UV/VIS sensor may be exploited. The topic of the PhD thesis work summarized here is the use of neural networks (...

We propose a new learning algorithm for multi Multi Layered Perceptron (MLP) architecture useful for complex classification. First, on the principle “divide and conquer”, a quantization is performed on the input space. One classical Neural Network (NN) is devoted to each subregion that results from this quantization. A cooperative learning procedure allows to train the NNs in such a way that NN...

The North Natuna Sea (NNS) conflict among some countries has been an important case in Indonesia. Related to the conflict, it will discuss Indonesia's future conditions faced with interests of China itself. NNS is a prolonged because security for itself, also related country which economic and territory China. From results predictive analysis both exploration, projection, forecasting, obtained ...

Neural networks (NNs) are increasingly applied in safety-critical systems such as autonomous vehicles. However, they fragile and often ill-behaved. Consequently, their behaviors should undergo rigorous guarantees before deployment practice. In this paper, we propose a set-boundary reachability method to investigate the safety verification problem of NNs from topological perspective. Given an NN...

Near Net Shape (NNS) manufacturing produces the engineering components closer to its final specification in single stage or minimum number of stages. ADI process technology is one of the NNS manufacturing technologies, currently being followed by the industries for manufacturing of engineering components at competitive price. Austempered Ductile Iron is a new class of engineering material which...

The backcalculation of pavement layer moduli from Falling Weight Deflectometer (FWD) measured surface deflections is a challenging task. It can also be formulated as a global optimization problem with the objective of finding the optimal pavement layer moduli values that minimize the error between measured and computed surface deflections. Over the years, several backcalculation methodologies h...

In this paper, the problem of stability criteria of neural networks (NNs) with two-additive time-varying delay compenents is investigated. The relationship between the time-varying delay and its lower and upper bounds is taken into account when estimating the upper bound of the derivative of Lyapunov functional. As a result, some improved delay stability criteria for NNs with two-additive time-...