In this work, we optimally solve the problem of multiplierless design of second-order Infinite Impulse Response filters with minimum number of adders. Given a frequency specification, we design a stable direct form filter with hardware-aware fixed-point coefficients that yielding minimal number of adders when replacing all the multiplications by bit shifts and additions. The coefficient design, quantization and implementation, typically conducted independently, are now gathered into one global optimization problem, modeled through integer linear programming and efficiently solved using generic solvers. We guarantee the frequency-domain specifications and stability, which together with optimal number of adders will significantly simplify design-space exploration for filter designers. The optimal filters are implemented within the FloPoCo IP core generator and synthesized for Field Programmable Gate Arrays. With respect to state-of-the-art three-step filter design methods, our one-step design approach achieves, on average, 42% reduction in the number of lookup tables and 21% improvement in delay.
相關內容
In this paper, we study how coded caching can be efficiently applied to multiple-input multiple-output (MIMO) communications. This is an extension to cache-aided multiple-input single-output (MISO) communications, where it is shown that with an $L$-antenna transmitter and coded caching gain $t$, a cumulative coded caching and spatial multiplexing gain of $t+L$ is achievable. We show that, interestingly, for MIMO setups with $G$-antenna receivers, a coded caching gain larger than MISO setups by a multiplicative factor of $G$ is possible, and the full coded caching and spatial multiplexing gain of $Gt+L$ is also achievable. Furthermore, we propose a novel algorithm for building low-subpacketization, high-performance MIMO coded caching schemes using a large class of existing MISO schemes.
Many software systems can be tuned for multiple objectives (e.g., faster runtime, less required memory, less network traffic or energy consumption, etc.). Optimizers built for different objectives suffer from "model disagreement"; i.e., they have different (or even opposite) insights and tactics on how to optimize a system. Model disagreement is rampant (at least for configuration problems). Yet prior to this paper, it has barely been explored. This paper shows that model disagreement can be mitigated via VEER, a one-dimensional approximation to the N-objective space. Since it is exploring a simpler goal space, VEER runs very fast (for eleven configuration problems). Even for our largest problem (with tens of thousands of possible configurations), VEER finds as good or better optimizations with zero model disagreements, three orders of magnitude faster (since its one-dimensional output no longer needs the sorting procedure). Based on the above, we recommend VEER as a very fast method to solve complex configuration problems, while at the same time avoiding model disagreement.
Block coordinate descent (BCD), also known as nonlinear Gauss-Seidel, is a simple iterative algorithm for nonconvex optimization that sequentially minimizes the objective function in each block coordinate while the other coordinates are held fixed. We propose a version of BCD that, for block multi-convex and smooth objective functions under constraints, is guaranteed to converge to the stationary points with worst-case rate of convergence of $O((\log n)^{2}/n)$ for $n$ iterations, and a bound of $O(\epsilon^{-1}(\log \epsilon^{-1})^{2})$ for the number of iterations to achieve an $\epsilon$-approximate stationary point. Furthermore, we show that these results continue to hold even when the convex sub-problems are inexactly solved if the optimality gaps are uniformly summable against initialization. A key idea is to restrict the parameter search within a diminishing radius to promote stability of iterates. As an application, we provide an alternating least squares algorithm with diminishing radius for nonnegative CP tensor decomposition that converges to the stationary points of the reconstruction error with the same robust worst-case convergence rate and complexity bounds. We also experimentally validate our results with both synthetic and real-world data and demonstrate that using auxiliary search radius restriction can in fact improve the rate of convergence.
We present two recursive strategy improvement algorithms for solving simple stochastic games. First we present an algorithm for solving SSGs of degree $d$ that uses at most $O\left(\left\lfloor(d+1)^2/2\right\rfloor^{n/2}\right)$ iterations, with $n$ the number of MAX vertices. Then, we focus on binary SSG and propose an algorithm that has complexity $O\left(\varphi^nPoly(N)\right)$ where $\varphi = (1 + \sqrt{5})/2$ is the golden ratio. To the best of our knowledge, this is the first deterministic strategy improvement algorithm that visits $2^{cn}$ strategies with $c < 1$.
An orthogonal representation of a graph $G$ over a field $\mathbb{F}$ is an assignment of a vector $u_v \in \mathbb{F}^t$ to every vertex $v$ of $G$, such that $\langle u_v,u_v \rangle \neq 0$ for every vertex $v$ and $\langle u_v,u_{v'} \rangle = 0$ whenever $v$ and $v'$ are adjacent in $G$. The locality of the orthogonal representation is the largest dimension of a subspace spanned by the vectors associated with a closed neighborhood in the graph. We introduce a novel graph parameter, called the local orthogonality dimension, defined for a given graph $G$ and a given field $\mathbb{F}$, as the smallest possible locality of an orthogonal representation of $G$ over $\mathbb{F}$. We investigate the usefulness of topological methods for proving lower bounds on the local orthogonality dimension. We prove that graphs for which topological methods imply a lower bound of $t$ on their chromatic number have local orthogonality dimension at least $\lceil t/2 \rceil +1$ over every field, strengthening a result of Simonyi and Tardos on the local chromatic number. We show that for certain graphs this lower bound is tight, whereas for others, the local orthogonality dimension over the reals is equal to the chromatic number. More generally, we prove that for every complement of a line graph, the local orthogonality dimension over $\mathbb{R}$ coincides with the chromatic number. This strengthens a recent result by Daneshpajouh, Meunier, and Mizrahi, who proved that the local and standard chromatic numbers of these graphs are equal. As another extension of their result, we prove that the local and standard chromatic numbers are equal for some additional graphs, from the family of Kneser graphs. We also show an $\mathsf{NP}$-hardness result for the local orthogonality dimension and present an application of this graph parameter to the index coding problem from information theory.
We consider expected risk minimization when the range of the estimator is required to be nonnegative, motivated by the settings of maximum likelihood estimation (MLE) and trajectory optimization. To facilitate nonlinear interpolation, we hypothesize that search is conducted over a Reproducing Kernel Hilbert Space (RKHS). To solve it, we develop first and second-order variants of stochastic mirror descent employing (i) pseudo-gradients and (ii) complexity-reducing projections. Compressive projection in first-order scheme is executed via kernel orthogonal matching pursuit (KOMP), and overcome the fact that the vanilla RKHS parameterization grows unbounded with time. Moreover, pseudo-gradients are needed when stochastic estimates of the gradient of the expected cost are only computable up to some numerical errors, which arise in, e.g., integral approximations. The second-order scheme develops a Hessian inverse approximation via recursively averaged pseudo-gradient outer products. For the first-order scheme, we establish tradeoffs between accuracy of convergence in mean and the projection budget parameter under constant step-size and compression budget are established, as well as non-asymptotic bounds on the model complexity. Analogous convergence results are established for the second-order scheme under an additional eigenvalue decay condition on the Hessian of the optimal RKHS element. Experiments demonstrate favorable performance on inhomogeneous Poisson Process intensity estimation in practice.
We consider the optimization of a two-hop relay network based on an amplify-and-forward Multiple-Input Multiple-Output (MIMO) relay. The relay is assumed to derive the output signal by a Relay Transform Matrix (RTM) applied to the input signal. Assuming perfect channel state information about the network at the relay, the RTM is optimized according to two different criteria: {\bf\em i)} network capacity; {\bf\em ii)} network capacity based on Orthogonal Space--Time Block Codes. The two assumptions have been addressed in part in the literature. The optimization problem is reduced to a manageable convex form, whose KKT equations are explicitly solved. Then, a parametric solution is given, which yields the power constraint and the capacity achieved with uncorrelated transmitted data as functions of a positive indeterminate. The solution for a given average power constraint at the relay is amenable to a \emph{water-filling-like} algorithm, and extends earlier literature results addressing the case without the direct link. Simulation results are reported concerning a Rayleigh relay network and the role of the direct link SNR is precisely assessed.
A new converse bound is presented for the two-user multiple-access channel under the average probability of error constraint. This bound shows that for most channels of interest, the second-order coding rate -- that is, the difference between the best achievable rates and the asymptotic capacity region as a function of blocklength $n$ with fixed probability of error -- is $O(1/\sqrt{n})$ bits per channel use. The principal tool behind this converse proof is a new measure of dependence between two random variables called wringing dependence, as it is inspired by Ahlswede's wringing technique. The $O(1/\sqrt{n})$ gap is shown to hold for any channel satisfying certain regularity conditions, which includes all discrete-memoryless channels and the Gaussian multiple-access channel. Exact upper bounds as a function of the probability of error are proved for the coefficient in the $O(1/\sqrt{n})$ term, although for most channels they do not match existing achievable bounds.
We give a short proof of a stronger form of the Johansson-Molloy theorem which relies only on the first moment method. The proof adapts a clever counting argument developed by Rosenfeld in the context of non-repetitive colourings. We then extend that result to graphs where each neighbourhood has bounded density, which improves a recent result from Davies et al. Focusing on tightening the number of colours, we obtain the best known upper bound for the chromatic number of triangle-free graphs of maximum degree $\Delta \ge 224$.
Multiplying matrices is among the most fundamental and compute-intensive operations in machine learning. Consequently, there has been significant work on efficiently approximating matrix multiplies. We introduce a learning-based algorithm for this task that greatly outperforms existing methods. Experiments using hundreds of matrices from diverse domains show that it often runs $100\times$ faster than exact matrix products and $10\times$ faster than current approximate methods. In the common case that one matrix is known ahead of time, our method also has the interesting property that it requires zero multiply-adds. These results suggest that a mixture of hashing, averaging, and byte shuffling$-$the core operations of our method$-$could be a more promising building block for machine learning than the sparsified, factorized, and/or scalar quantized matrix products that have recently been the focus of substantial research and hardware investment.
小貼士
登錄享
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191