We report on a fully closed-form approximation of the GN model for Nyquist WDM systems that extends the range of applicability of previously available formulas to values of fiber loss and span loss that were not previously covered. In particular, so far, a closed-form formula was available for zero loss, and another for span loss of at least about 7 dB. The new formula is accurate over any fiber loss and any span loss value. The interest for this new formula is varied, ranging from hollow-core fibers to improving modeling of standard systems that present lumped loss along installed spans (such as splices or connectors).
The paper Zhao et al. (2015) shows that mean-CVaR-skewness portfolio optimization problems based on asymetric Laplace (AL) distributions can be transformed into quadratic optimization problems under which closed form solutions can be found. In this note, we show that such result also holds for mean-risk-skewness portfolio optimization problems when the underlying distribution is a larger class of normal mean-variance mixture (NMVM) models than the class of AL distributions. We then study the value at risk (VaR) and conditional value at risk (CVaR) risk measures on portfolios of returns with NMVM distributions. They have closed form expressions for portfolios of normal and more generally elliptically distributed returns as discussed in Rockafellar & Uryasev (2000) and in Landsman & Valdez (2003). When the returns have general NMVM distributions, these risk measures do not give closed form expressions. In this note, we give approximate closed form expressions for VaR and CVaR of portfolios of returns with NMVM distributions. Numerical tests show that our closed form formulas give accurate values for VaR and CVaR and shortens the computational time for portfolio optimization problems associated with VaR and CVaR considerably.
Multi-talker conversational speech processing has drawn many interests for various applications such as meeting transcription. Speech separation is often required to handle overlapped speech that is commonly observed in conversation. Although the original utterancelevel permutation invariant training-based continuous speech separation approach has proven to be effective in various conditions, it lacks the ability to leverage the long-span relationship of utterances and is computationally inefficient due to the highly overlapped sliding windows. To overcome these drawbacks, we propose a novel training scheme named Group-PIT, which allows direct training of the speech separation models on the long-form speech with a low computational cost for label assignment. Two different speech separation approaches with Group-PIT are explored, including direct long-span speech separation and short-span speech separation with long-span tracking. The experiments on the simulated meeting-style data demonstrate the effectiveness of our proposed approaches, especially in dealing with a very long speech input.
The relevance of polynomial formula classes to deductive efficiency motivated their search, and currently, a great number of such classes is known. Nonetheless, they have been exclusively sought in the setting of clausal form and propositional logic, which is of course expressively limiting for real applications. As a consequence, a first polynomial propositional class in non-clausal (NC) form has recently been proposed. Along these lines and towards making NC tractability applicable beyond propositional logic, firstly, we define the Regular many-valued Horn Non-Clausal class, or RH, obtained by suitably amalgamating both regular classes: Horn and NC. Secondly, we demonstrate that the relationship between (1) RH and the regular Horn class is that syntactically RH subsumes the Horn class but that both classes are equivalent semantically; and between (2) RH and the regular non-clausal class is that RH contains all NC formulas whose clausal form is Horn. Thirdly, we define Regular Non-Clausal Unit-Resolution, or RUR-NC , and prove both that it is complete for RH and that checks its satisfiability in polynomial time. The latter fact shows that our intended goal is reached since RH is many-valued, non-clausal and tractable. As RH and RUR-NC are, both, basic in the DPLL scheme, the most efficient in propositional logic, and can be extended to some other non-classical logics, we argue that they pave the way for efficient non-clausal DPLL-based approximate reasoning.
Reinforcement Learning faces an important challenge in partial observable environments that has long-term dependencies. In order to learn in an ambiguous environment, an agent has to keep previous perceptions in a memory. Earlier memory based approaches use a fixed method to determine what to keep in the memory, which limits them to certain problems. In this study, we follow the idea of giving the control of the memory to the agent by allowing it to have memory-changing actions. This learning mechanism is supported by an intrinsic motivation to memorize rare observations that can help the agent to disambiguate its state in the environment. Our approach is experimented and analyzed on several partial observable tasks with long-term dependencies and compared with other memory based methods.
Question answering (QA) is a high-level ability of natural language processing. Most extractive ma-chine reading comprehension models focus on factoid questions (e.g., who, when, where) and restrict the output answer as a short and continuous span in the original passage. However, in real-world scenarios, many questions are non-factoid (e.g., how, why) and their answers are organized in the list format that contains multiple non-contiguous spans. Naturally, existing extractive models are by design unable to answer such questions. To address this issue, this paper proposes ListReader, a neural ex-tractive QA model for list-form answer. In addition to learning the alignment between the question and content, we introduce a heterogeneous graph neural network to explicitly capture the associations among candidate segments. Moreover, our model adopts a co-extraction setting that can extract either span- or sentence-level answers, allowing better applicability. Two large-scale datasets of different languages are constructed to support this study. Experimental results show that our model considerably outperforms various strong baselines. Further discussions provide an intuitive understanding of how our model works and where the performance gain comes from.
We devise a multimodal conversation system for dialogue utterances composed of text, image or both modalities. We leverage Auxiliary UnsuperviseD vIsual and TExtual Data (AUDITED). To improve the performance of text-based task, we utilize translations of target sentences from English to French to form the assisted supervision. For the image-based task, we employ the DeepFashion dataset in which we seek nearest neighbor images of positive and negative target images of the MMD data. These nearest neighbors form the nearest neighbor embedding providing an external context for target images. We form two methods to create neighbor embedding vectors, namely Neighbor Embedding by Hard Assignment (NEHA) and Neighbor Embedding by Soft Assignment (NESA) which generate context subspaces per target image. Subsequently, these subspaces are learnt by our pipeline as a context for the target data. We also propose a discriminator which switches between the image- and text-based tasks. We show improvements over baselines on the large-scale Multimodal Dialogue Dataset (MMD) and SIMMC.
We make an attempt to understanding convolutional neural network by exploring the relationship between (deep) convolutional neural networks and Volterra convolutions. We propose a novel approach to explain and study the overall characteristics of neural networks without being disturbed by the horribly complex architectures. Specifically, we attempt to convert the basic structures of a convolutional neural network (CNN) and their combinations to the form of Volterra convolutions. The results show that most of convolutional neural networks can be approximated in the form of Volterra convolution, where the approximated proxy kernels preserve the characteristics of the original network. Analyzing these proxy kernels may give valuable insight about the original network. Base on this setup, we presented methods to approximating the order-zero and order-one proxy kernels, and verified the correctness and effectiveness of our results.
WiFi human sensing has become increasingly attractive in enabling emerging human-computer interaction applications. The corresponding technique has gradually evolved from the classification of multiple activity types to more fine-grained tracking of 3D human poses. However, existing WiFi-based 3D human pose tracking is limited to a set of predefined activities. In this work, we present Winect, a 3D human pose tracking system for free-form activity using commodity WiFi devices. Our system tracks free-form activity by estimating a 3D skeleton pose that consists of a set of joints of the human body. In particular, we combine signal separation and joint movement modeling to achieve free-form activity tracking. Our system first identifies the moving limbs by leveraging the two-dimensional angle of arrival of the signals reflected off the human body and separates the entangled signals for each limb. Then, it tracks each limb and constructs a 3D skeleton of the body by modeling the inherent relationship between the movements of the limb and the corresponding joints. Our evaluation results show that Winect is environment-independent and achieves centimeter-level accuracy for free-form activity tracking under various challenging environments including the none-line-of-sight (NLoS) scenarios.
We develop a weak-form sparse identification method for interacting particle systems (IPS) with the primary goals of reducing computational complexity for large particle number $N$ and offering robustness to either intrinsic or extrinsic noise. In particular, we use concepts from mean-field theory of IPS in combination with the weak-form sparse identification of nonlinear dynamics algorithm (WSINDy) to provide a fast and reliable system identification scheme for recovering the governing stochastic differential equations for an IPS when the number of particles per experiment $N$ is on the order of several thousand and the number of experiments $M$ is less than 100. This is in contrast to existing work showing that system identification for $N$ less than 100 and $M$ on the order of several thousand is feasible using strong-form methods. We prove that under some standard regularity assumptions the scheme converges with rate $\mathcal{O}(N^{-1/2})$ in the ordinary least squares setting and we demonstrate the convergence rate numerically on several systems in one and two spatial dimensions. Our examples include a canonical problem from homogenization theory (as a first step towards learning coarse-grained models), the dynamics of an attractive-repulsive swarm, and the IPS description of the parabolic-elliptic Keller-Segel model for chemotaxis.
This paper investigates the performance, in terms of choice probabilities and correlations, of existing and new specifications of closed-form route choice models with flexible correlation patterns, namely the Link Nested Logit (LNL), the Paired Combinatorial Logit (PCL) and the more recent Combination of Nested Logit (CoNL) models. Following a consolidated track in the literature, choice probabilities and correlations of the Multinomial Probit (MNP) model by (Daganzo and Sheffi, 1977) are taken as target. Laboratory experiments on small/medium-size networks are illustrated, also leveraging a procedure for practical calculation of correlations of any GEV models, proposed by (Marzano 2014). Results show that models with inherent limitations in the coverage of the domain of feasible correlations yield unsatisfactory performance, whilst the specifications of the CoNL proposed in the paper appear the best in fitting both MNP correlations and probabilities. Performance of the models are appreciably ameliorated by introducing lower bounds to the nesting parameters. Overall, the paper provides guidance for the practical application of tested models.
Weighted First-Order Model Counting (WFOMC) computes the weighted sum of the models of a first-order logic theory on a given finite domain. First-Order Logic theories that admit polynomial-time WFOMC w.r.t domain cardinality are called domain liftable. We introduce the concept of lifted interpretations as a tool for formulating closed-forms for WFOMC. Using lifted interpretations, we reconstruct the closed-form formula for polynomial-time FOMC in the universally quantified fragment of FO2, earlier proposed by Beame et al. We then expand this closed-form to incorporate cardinality constraints, existential quantifiers, and counting quantifiers (a.k.a C2) without losing domain-liftability. Finally, we show that the obtained closed-form motivates a natural definition of a family of weight functions strictly larger than symmetric weight functions.
We propose a novel framework to evaluate the robustness of transformer-based form field extraction methods via form attacks. We introduce 14 novel form transformations to evaluate the vulnerability of the state-of-the-art field extractors against form attacks from both OCR level and form level, including OCR location/order rearrangement, form background manipulation and form field-value augmentation. We conduct robustness evaluation using real invoices and receipts, and perform comprehensive research analysis. Experimental results suggest that the evaluated models are very susceptible to form perturbations such as the variation of field-values (~15% drop in F1 score), the disarrangement of input text order(~15% drop in F1 score) and the disruption of the neighboring words of field-values(~10% drop in F1 score). Guided by the analysis, we make recommendations to improve the design of field extractors and the process of data collection.
End-to-end models have achieved state-of-the-art results on several automatic speech recognition tasks. However, they perform poorly when evaluated on long-form data, e.g., minutes long conversational telephony audio. One reason the model fails on long-form speech is that it has only seen short utterances during training. In this paper we study the effect of training utterance length on the word error rate (WER) for RNN-transducer (RNN-T) model. We compare two widely used training objectives, log loss (or RNN-T loss) and minimum word error rate (MWER) loss. We conduct experiments on telephony datasets in four languages. Our experiments show that for both losses, the WER on long-form speech reduces substantially as the training utterance length increases. The average relative WER gain is 15.7% for log loss and 8.8% for MWER loss. When training on short utterances, MWER loss leads to a lower WER than the log loss. Such difference between the two losses diminishes when the input length increases.
We consider the fundamental problem of sampling the optimal transport coupling between given source and target distributions. In certain cases, the optimal transport plan takes the form of a one-to-one mapping from the source support to the target support, but learning or even approximating such a map is computationally challenging for large and high-dimensional datasets due to the high cost of linear programming routines and an intrinsic curse of dimensionality. We study instead the Sinkhorn problem, a regularized form of optimal transport whose solutions are couplings between the source and the target distribution. We introduce a novel framework for learning the Sinkhorn coupling between two distributions in the form of a score-based generative model. Conditioned on source data, our procedure iterates Langevin Dynamics to sample target data according to the regularized optimal coupling. Key to this approach is a neural network parametrization of the Sinkhorn problem, and we prove convergence of gradient descent with respect to network parameters in this formulation. We demonstrate its empirical success on a variety of large scale optimal transport tasks.
The present paper treats the identification of nonlinear dynamical systems using Koopman-based deep state-space encoders. Through this method, the usual drawback of needing to choose a dictionary of lifting functions a priori is circumvented. The encoder represents the lifting function to the space where the dynamics are linearly propagated using the Koopman operator. An input-affine formulation is considered for the lifted model structure and we address both full and partial state availability. The approach is implemented using the the deepSI toolbox in Python. To lower the computational need of the simulation error-based training, the data is split into subsections where multi-step prediction errors are calculated independently. This formulation allows for efficient batch optimization of the network parameters and, at the same time, excellent long term prediction capabilities of the obtained models. The performance of the approach is illustrated by nonlinear benchmark examples.
Numerous studies of the tumor microenvironment, interspecies xenografting, and limb regeneration suggest the existence of a Tissue Spatial Code (TSC) that controls tissue structure in a quasi-epigenetic fashion. The term 'epigenetic' is an inadequate label for the concept that encompasses the TSC, because this information does not act upon DNA or chromatin via post-translational modifications (i.e. DNA methylation, histone acetylation). A broader term is needed to capture the diversity of three-dimensional (3D) spatial codes in biology. One such term is Heritable Nongenetic Information (HNI, pronounced 'honey'), which encompasses the TSC. The term 'heritable' is appropriate, because this information is passed onto offspring, otherwise it would have disappeared during evolution. Another reason for the heritability of HNI is that the spatial information observed in tissues is not reducible to the laws of physics, meaning structures like epithelial tubes or neural circuits do not spontaneously form in an aqueous solution; pre-existing physiological information in the microenvironment is necessary. In summary, HNI is defined as: heritable, instructional information that affects molecular-, cellular-, tissue-, or organ-level function without having to post-translationally modify DNA or chromatin in order to be inherited.
The paradigm of differentiable programming has significantly enhanced the scope of machine learning via the judicious use of gradient-based optimization. However, standard differentiable programming methods (such as autodiff) typically require that the machine learning models be differentiable, limiting their applicability. Our goal in this paper is to use a new, principled approach to extend gradient-based optimization to functions well modeled by splines, which encompass a large family of piecewise polynomial models. We derive the form of the (weak) Jacobian of such functions and show that it exhibits a block-sparse structure that can be computed implicitly and efficiently. Overall, we show that leveraging this redesigned Jacobian in the form of a differentiable "layer" in predictive models leads to improved performance in diverse applications such as image segmentation, 3D point cloud reconstruction, and finite element analysis.
Compared to point estimates calculated by standard neural networks, Bayesian neural networks (BNN) provide probability distributions over the output predictions and model parameters, i.e., the weights. Training the weight distribution of a BNN, however, is more involved due to the intractability of the underlying Bayesian inference problem and thus, requires efficient approximations. In this paper, we propose a novel approach for BNN learning via closed-form Bayesian inference. For this purpose, the calculation of the predictive distribution of the output and the update of the weight distribution are treated as Bayesian filtering and smoothing problems, where the weights are modeled as Gaussian random variables. This allows closed-form expressions for training the network's parameters in a sequential/online fashion without gradient descent. We demonstrate our method on several UCI datasets and compare it to the state of the art.
Predicting fading channels is a classical problem with a vast array of applications, including as an enabler of artificial intelligence (AI)-based proactive resource allocation for cellular networks. Under the assumption that the fading channel follows a stationary complex Gaussian process, as for Rayleigh and Rician fading models, the optimal predictor is linear, and it can be directly computed from the Doppler spectrum via standard linear minimum mean squared error (LMMSE) estimation. However, in practice, the Doppler spectrum is unknown, and the predictor has only access to a limited time series of estimated channels. This paper proposes to leverage meta-learning in order to mitigate the requirements in terms of training data for channel fading prediction. Specifically, it first develops an offline low-complexity solution based on linear filtering via a meta-trained quadratic regularization. Then, an online method is proposed based on gradient descent and equilibrium propagation (EP). Numerical results demonstrate the advantages of the proposed approach, showing its capacity to approach the genie-aided LMMSE solution with a small number of training data points.