We develop a stochastic whole-brain and body simulator of the nematode roundworm Caenorhabditis elegans (C. elegans) and show that it is sufficiently regularizing to allow imputation of latent membrane potentials from partial calcium fluorescence imaging observations. This is the first attempt we know of to "complete the circle," where an anatomically grounded whole-connectome simulator is used to impute a time-varying "brain" state at single-cell fidelity from covariates that are measurable in practice. The sequential Monte Carlo (SMC) method we employ not only enables imputation of said latent states but also presents a strategy for learning simulator parameters via variational optimization of the noisy model evidence approximation provided by SMC. Our imputation and parameter estimation experiments were conducted on distributed systems using novel implementations of the aforementioned techniques applied to synthetic data of dimension and type representative of that which are measured in laboratories currently.

--The location-based social network, FourSquare, helps us to understand a city's mass human mobility. It provides data that characterises the volume of movements across regions and Places of Interests (POIs) to explore the crime dynamics of a city. T o fully exploit human movement into crime analysis, we propose the region risk factor which combines monthly aggregated crime and human movement of a region across different time intervals. We then derive a number of features using the region risk factor and conduct extensive experiments with real world data in multiple cities that verify the effectiveness of these features. One of the basic demands for every person in society is a safe and secure living space.

Transportation systems can be conceptualized as an instrument of spreading people and resources over the territory, playing an important role in developing sustainable cities. The current rationale of transport provision is based on population demand, disregarding land use and socioeconomic information. To meet the challenge to promote a more equitable resource distribution, this work aims at identifying and describing patterns of urban services supply, their accessibility, and household income. By using a multidimensional approach, the spatial inequalities of a large city of the global south reveal that the low-income population has low access mainly to hospitals and cultural centers. A low-income group presents an intermediate level of accessibility to public schools and sports centers, evidencing the diverse condition of citizens in the peripheries. These complex outcomes generated by the interaction of land use and public transportation emphasize the importance of comprehensive methodological approaches to support decisions of urban projects, plans and programs. Reducing spatial inequalities, especially providing services for deprived groups, is fundamental to promote the sustainable use of resources and optimize the daily commuting.

The problem of ranking is a multi-billion dollar problem. In this paper we present an overview of several production quality ranking systems. We show that due to conflicting goals of employing the most effective machine learning models and responding to users in real time, ranking systems have evolved into a system of systems, where each subsystem can be viewed as a component layer. We view these layers as being data processing, representation learning, candidate selection and online inference. Each layer employs different algorithms and tools, with every end-to-end ranking system spanning multiple architectures. Our goal is to familiarize the general audience with a working knowledge of ranking at scale, the tools and algorithms employed and the challenges introduced by adopting a layered approach.

The computational cost of training with softmax cross entropy loss grows linearly with the number of classes. For the settings where a large number of classes are involved, a common method to speed up training is to sample a subset of classes and utilize an estimate of the gradient based on these classes, known as the sampled softmax method. However, the sampled softmax provides a biased estimate of the gradient unless the samples are drawn from the exact softmax distribution, which is again expensive to compute. Therefore, a widely employed practical approach (without theoretical justification) involves sampling from a simpler distribution in the hope of approximating the exact softmax distribution. In this paper, we develop the first theoretical understanding of the role that different sampling distributions play in determining the quality of sampled softmax. Motivated by our analysis and the work on kernel-based sampling, we propose the Random Fourier Softmax (RF-softmax) method that utilizes the powerful Random Fourier features to enable more efficient and accurate sampling from the (approximate) softmax distribution. We show that RF-softmax leads to low bias in estimation in terms of both the full softmax distribution and the full softmax gradient. Furthermore, the cost of RF-softmax scales only logarithmically with the number of classes.

While stochastic gradient descent (SGD) and variants have been surprisingly successful for training deep nets, several aspects of the optimization dynamics and generalization are still not well understood. In this paper, we present new empirical observations and theoretical results on both the optimization dynamics and generalization behavior of SGD for deep nets based on the Hessian of the training loss and associated quantities. We consider three specific research questions: (1) what is the relationship between the Hessian of the loss and the second moment of stochastic gradients (SGs)? (2) how can we characterize the stochastic optimization dynamics of SGD with fixed and adaptive step sizes and diagonal pre-conditioning based on the first and second moments of SGs? and (3) how can we characterize a scale-invariant generalization bound of deep nets based on the Hessian of the loss, which by itself is not scale invariant? We shed light on these three questions using theoretical results supported by extensive empirical observations, with experiments on synthetic data, MNIST, and CIFAR-10, with different batch sizes, and with different difficulty levels by synthetically adding random labels.

We consider a decentralized learning problem, where a set of computing nodes aim at solving a non-convex optimization problem collaboratively. It is well-known that decentralized optimization schemes face two major system bottlenecks: stragglers' delay and communication overhead. In this paper, we tackle these bottlenecks by proposing a novel decentralized and gradient-based optimization algorithm named as QuanTimed-DSGD. Our algorithm stands on two main ideas: (i) we impose a deadline on the local gradient computations of each node at each iteration of the algorithm, and (ii) the nodes exchange quantized versions of their local models. The first idea robustifies to straggling nodes and the second alleviates communication efficiency. The key technical contribution of our work is to prove that with non-vanishing noises for quantization and stochastic gradients, the proposed method exactly converges to the global optimal for convex loss functions, and finds a first-order stationary point in non-convex scenarios. Our numerical evaluations of the QuanTimed-DSGD on training benchmark datasets, MNIST and CIFAR-10, demonstrate speedups of up to 3x in run-time, compared to state-of-the-art decentralized optimization methods.

The importance weighted autoencoder (IWAE) (Burda et al., 2016) and reweighted wake-sleep (RWS) algorithm (Bornschein and Bengio, 2015) are popular approaches which employ multiple samples to achieve bias reductions compared to standard variational methods. However, their relationship has hitherto been unclear. We introduce a simple, unified framework for multi-sample variational inference termed adaptive importance sampling for learning (AISLE) and show that it admits IWAE and RWS as special cases. Through a principled application of a variance-reduction technique from Tucker et al. (2019), we also show that the sticking-the-landing (STL) gradient from Roeder et al. (2017), which previously lacked theoretical justification, can be recovered as a special case of RWS (and hence of AISLE). In particular, this indicates that the breakdown of RWS -- but not of STL -- observed in Tucker et al. (2019) may not be attributable to the lack of a joint objective for the generative-model and inference-network parameters as previously conjectured. Finally, we argue that our adaptive-importance-sampling interpretation of variational inference leads to more natural and principled extensions to sequential Monte Carlo methods than the IWAE-type multi-sample objective interpretation.

Y adav and BodeMETHODOLOGY A graphical heuristic for reduction and partitioning of large datasets for scalable supervised training Sumedh Y adav 1* and Mathis Bode 2 * Correspondence: sumedhyadav.iitkgp@gmail.com 1 Gstech T echnology Pvt. Ltd., 415, 2nd Floor, 16th Cross Road, 17th Main Road, HSR Layout Sector 4, 560102, Bengaluru, India Full list of author information is available at the end of the article Abstract A scalable graphical method is presented for selecting, and partitioning datasets for the training phase of a classification task. For the heuristic, a clustering algorithm is required to get its computation cost in a reasonable proportion to the task itself. This step is proceeded by construction of an information graph of the underlying classification patterns using approximate nearest neighbor methods. The presented method constitutes of two approaches, one for reducing a given training set, and another for partitioning the selected/reduced set. The heuristic targets large datasets, since the primary goal is significant reduction in training computation run-time without compromising prediction accuracy . T est results show that both approaches significantly speedup the training task when compared against that of state-of-the-art shrinking heuristic available in LIBSVM. Furthermore, the approaches closely follow or even outperform in prediction accuracy . A network design is also presented for the partitioning based distributed training formulation. Added speedup in training run-time is observed when compared to that of serial implementation of the approaches. Keywords: training set selection; machine learning; large datasets; distributed machine learning; classification; graph coarsening objective; network architecture design Introduction Two decades earlier, some of the most seminal works in machine learning were done on training set selection [1, 2] under the banner of relevance reasoning. However, the better part of recent works have been exclusively towards feature selection [3, 4]. With increased processing power, run time of training is feasible even for datasets erstwhile considered large. Additionally, dimensionality ( d) dominates dataset size ( n) in the algorithmic complexities of learning algorithms. In the training phase, less data points mean fewer generalization guarantees, however, as we are moving in the era of big data, even the fastest classification algorithms are taking unfeasible time to train models. When data sources are abundant, it is befitting to separate data based on relevance to the learning task. This has led to a renewed interest in the once famous problem statement of relevance reasoning [5, 6]. Reasoning on relevance to get improved scalability of classification algorithms is currently explored on graphical/network data [7], and learned models [8]. One research area where training set selection has been given attention to is support vector machines (SVM).

In this work, we study constrained clustering, where constraints are utilized to guide the clustering process. In existing works, two categories of constraints have been widely explored, namely pairwise and cardinality constraints. Pairwise constraints enforce the cluster labels of two instances to be the same (must-link constraints) or different (cannot-link constraints). Cardinality constraints encourage cluster sizes to satisfy a user-specified distribution. However, most existing constrained clustering models can only utilize one category of constraints at a time. In this paper, we enforce the above two categories into a unified clustering model starting with the integer program formulation of the standard K-means. As these two categories provide useful information at different levels, utilizing both of them is expected to allow for better clustering performance. However, the optimization is difficult due to the binary and quadratic constraints in the proposed unified formulation. To alleviate this difficulty, we utilize two techniques: equivalently replacing the binary constraints by the intersection of two continuous constraints; the other is transforming the quadratic constraints into bi-linear constraints by introducing extra variables. Then we derive an equivalent continuous reformulation with simple constraints, which can be efficiently solved by Alternating Direction Method of Multipliers (ADMM) algorithm. Extensive experiments on both synthetic and real data demonstrate: (1) when utilizing a single category of constraint, the proposed model is superior to or competitive with state-of-the-art constrained clustering models, and (2) when utilizing both categories of constraints jointly, the proposed model shows better performance than the case of the single category.