In this work, we experimentally demonstrate two key building blocks for realizing Binary/Ternary Neural Networks (BNNs/TNNs): (i) 130 nm CMOS based sigmoidal neurons and (ii) HfO$_{2}$ multi-level (MLC) OxRAM-synaptic blocks. An optimized vector matrix multiplication (VMM) programming scheme that utilizes the is also present...

We propose an efficient implementation of an interior tomography reconstruction method based on a known subregion. This method iteratively refines a reconstruction, aiming at reducing the local tomography artifacts. To cope with the ever increasing data volumes, this method is highly optimized on two aspects: firstly, the problem is reformulated to reduce the number of variables, and secondly, ...

This research presents a distributed and formula-based bilateration algorithm that can be used to provide initial set of locations. In this scheme each node uses distance estimates to anchors to solve a set of circle-circle intersection (CCI) problems, solved through a purely geometric formulation. The resulting CCIs are processed to pick those that cluster together and then take the average to...

The characteristics of users' browsing behaviors on websites can be used to analyze system performance as well as network communication, understand users’ reaction and motivation, and build adaptive websites. However, the motivation, requirement and experience of users may dynamically change, which cause difficulty in exactly refining a stable behavior pattern and describing their shifted inter...

The swing-twist decomposition is a standard routine in motion planning for humanoid limbs. In this paper the decomposition formulas are derived and discussed in terms of Clifford algebra. With the decomposition one can express an arbitrary spinor as a product of a twist-free spinor and a swingfree spinor (or vice-versa) in 3-dimensional Euclidean space. It is shown that in the derived decomposi...

This paper evaluates two methods for generating power or energy optimized Digital Signal Processing (DSP) transforms. Both approaches integrate power models into the SPIRAL code generator. SPIRAL decomposes a transform into an algorithmic formula representation, which is then implemented in C and timed. The timing data is used as feedback to control the parameters of the next algorithm generate...

Two methods are investigated for the time-domain simulation of functions and dynamical systems of Bessel type, involved in wave propagation (see e.g. [1], [8], [2]). Both are based on complex analysis and lead to finitedimensional approximations. The first method relies on optimized parametric contours and provides asymptotic convergence rates. The second is based on cuts and integral represent...

In this paper, we present a simple empirical model for calculating gain and excess noise in heterojunction GaAs/AlξGa1−ξAs APDs (0.3 ≤ ξ ≤ 0.6), without going through the relatively complicated and time consuming Monte Carlo simulation, commonly used for such devices. In this model, we present a set of empirical formula which can predict a distribution function for ionization path length for a ...

A Sudoku puzzle can be regarded as a propositional SAT problem. Various encodings are known for encoding Sudoku as a Conjunctive Normal Form(CNF) formula. Using these encodings for large Sudoku puzzles, however, generates too many clauses, which impede the performance of state-of-the-art SAT solvers. This paper presents an optimized CNF encoding in order to deal with large instances of Sudoku p...

In this article we first show how the functional and the optimized functional translation from modal logic to many-sorted first-order logic can be naturally extended to the hybrid language H(@, ↓). The translation is correct not only when reasoning over the class of all models, but for any firstorder definable class. We then show that sorts can be safely removed (i.e., without affecting the sat...