We will review some of the major results in random graphs and some of the more challenging open problems. We will cover algorithmic and structural questions. We will touch on newer models, including those related to the WWW.

Random graphs

Matrix Factorization Techniques for Recommender Systems

The state-of-the-art models for medical image segmentation are variants of U-Net and fully convolutional networks (FCN). Despite their success, these models have two limitations: (1) their optimal depth is apriori unknown, requiring extensive architecture search or inefficient ensemble of models of varying depths; and (2) their skip connections impose an unnecessarily restrictive fusion scheme, forcing aggregation only at the same-scale feature maps of the encoder and decoder sub-networks. To overcome these two limitations, we propose UNet++, a new neural architecture for semantic and instance segmentation, by (1) alleviating the unknown network depth with an efficient ensemble of U-Nets of varying depths, which partially share an encoder and co-learn simultaneously using deep supervision; (2) redesigning skip connections to aggregate features of varying semantic scales at the decoder sub-networks, leading to a highly flexible feature fusion scheme; and (3) devising a pruning scheme to accelerate the inference speed of UNet++. We have evaluated UNet++ using six different medical image segmentation datasets, covering multiple imaging modalities such as computed tomography (CT), magnetic resonance imaging (MRI), and electron microscopy (EM), and demonstrating that (1) UNet++ consistently outperforms the baseline models for the task of semantic segmentation across different datasets and backbone architectures; (2) UNet++ enhances segmentation quality of varying-size objects—an improvement over the fixed-depth U-Net; (3) Mask RCNN++ (Mask R-CNN with UNet++ design) outperforms the original Mask R-CNN for the task of instance segmentation; and (4) pruned UNet++ models achieve significant speedup while showing only modest performance degradation. Our implementation and pre-trained models are available at https://github.com/MrGiovanni/UNetPlusPlus .

UNet++: Redesigning Skip Connections to Exploit Multiscale Features in Image Segmentation

This paper proposes a way to incorporate the idea of spiking neurons into the area of membrane computing, and to this aim we introduce a class of neural-like P systems which we call spiking neural P systems (in short, SN P systems). In these devices, the time (when the neurons fire and/or spike) plays an essential role. For instance, the result of a computation is the time between the moments when a specified neuron spikes. Seen as number computing devices, SN P systems are shown to be computationally complete (both in the generating and accepting modes, in the latter case also when restricting to deterministic systems). If the number of spikes present in the system is bounded, then the power of SN P systems falls drastically, and we get a characterization of semilinear sets. A series of research topics and open problems are formulated.

/pdf/spiking-neural-p-systems-63ihtdu60m.pdf

Spiking Neural P Systems

Brass AL, Dykxhoorn DM, Benita Y, et al. Science . 2008;319(5865):921–926 PURPOSE OF THE STUDY. The HIV genome encodes only 15 proteins and, therefore, must use multiple host-cell collaborators for successful replication and transmission. Required host-derived proteins include CD4 as the primary virus receptor and chemokine receptors as coreceptors. This study identified multiple other host proteins required for HIV activity. METHODS. Human cells known to be susceptible to HIV were exposed in vitro to HIV. Using small interfering RNAs able to inhibit each known gene in the human genome 1 at a time, the investigators tested whether HIV could establish an infection and copy itself. HIV dependence on >21 000 human genes was examined. RESULTS. More than 250 human genes were identified to be required for efficient HIV replication. Termed “HIV-dependency factors,” the products of these genes are known to participate in a broad array of cellular functions and implicate unsuspected pathways in the virus life cycle. CONCLUSIONS. The extraordinary dependence of HIV on human host proteins for efficient transmission and replication provides many new potential targets for antiretroviral therapy. REVIEWER COMMENTS. An example of targeting host proteins is the use of chemokine receptor 5 (CCR5) inhibitors. Many people with CCR5 deficiency are very resistant to HIV infection yet have limited if any clinical consequences. Maraviroc CCR5 inhibitor is approved for treatment for HIV infection. This study identified many more such potential targets.

Identification of Host Proteins Required for HIV Infection Through a Functional Genomic Screen

An incremental approach to construction of biomolecular algorithms solving intractable problems is presented. The core idea is to build gradually the space of candidate solutions and remove invalid solutions as soon as possible. We demonstrate two examples of this strategy: a P system with replication and inhibitors for solving the Maximum Clique Problem for a graph, and an incremental DNA algorithm for the same problem inspired by the membrane solution. The DNA implementation is based on the parallel filtering DNA model featuring error-resistance of the employed operations. The algorithm is compared with two standard papers that addressed the same problem and its DNA implementation in the past. The comparison is carried out on the basis of a series of computational and physical parameters. The incremental algorithm features a dramatically lower cost in terms of time, the number and size of DNA strands, together with a high error-resistance. A probabilistic analysis shows that physical parameters (volume of the DNA pool, concentration of the solution-encoding strands) and error-resistance of the algorithm should allow to process in vitro instances of graphs with hundreds to thousands of vertices.

Towards a robust biocomputing solution of intractable problems

The paper summarizes recent knowledge about computational power of spiking neural P systems and presents a sequence of new more general results. The concepts of recognizer SN P systems and of uniform families of SN P systems provide a formal framework for this study. We establish the relation of computational power of spiking neural P systems with various limitations to standard complexity classes like P, NP, PSPACE and P/poly.

On the power of families of recognizer spiking neural p systems

We present a biomolecular probabilistic model driven by the action of a DNA toolbox made of a set of DNA templates and enzymes that is able to perform Bayesian inference. The model will take single-stranded DNA as input data, representing the presence or absence of a specific molecular signal the evidence. The program logic uses different DNA templates and their relative concentration ratios to encode the prior probability of a disease and the conditional probability of a signal given the disease. When the input and program molecules interact, an enzyme-driven cascade of reactions DNA polymerase extension, nicking and degradation is triggered, producing a different pair of single-stranded DNA species. Once the system reaches equilibrium, the ratio between the output species will represent the application of Bayes' law: the conditional probability of the disease given the signal. In other words, a qualitative diagnosis plus a quantitative degree of belief in that diagnosis. Thanks to the inherent amplification capability of this DNA toolbox, the resulting system will be able to to scale up with longer cascades and thus more input signals a Bayesian biosensor that we designed previously.

/pdf/probabilistic-reasoning-with-an-enzyme-driven-dna-device-4mn2a9ujto.pdf

Probabilistic Reasoning with an Enzyme-Driven DNA Device

The aim of this project is to contribute both to the progress of Natural Computing and Systems Biology, using and developing: (1) the distributed computational model inspired in the living cell called “P System” (or membrane computing), (2) biomolecular computing, and (3) microfluidic systems. The contribution of this project to these scientific areas can be split into the following objectives: 1. Modelling of cellular and neurobiological processes developing new P systems. 2. Description and mathematical formalization of biomolecular processes using automata and formal language theory. 3. Resolution of computational problems with biological motivation using distributed molecular automata and microfluidic systems.

/pdf/desarrollo-de-aplicaciones-de-la-computacion-celular-con-3l2b7hvpq9.pdf

Desarrollo de Aplicaciones de la Computación Celular con Membranas y la Computación Biomolecular en la Biología TIN2006-15595

Individual-based models are complex and they have usually an elevated number of input parameters which must be tuned for reproducing the observed population data or the experimental results as accurately as possible. Thus, one of the weakest points of this modelling approach lies on the fact that rarely the modeler has the enough information about the correct values or even the acceptable range for the input parameters. Consequently, several parameter combinations must be tried to find an acceptable set of input factors minimizing the deviations of simulated and the reference dataset. In practice, most of times, it is computationally unfeasible to traverse the complete search space trying all every possible combination to find the best of set of parameters. That is precisely an instance of a combinatorial problem which is suitable for being solved by metaheuristics and evolutionary computation techniques. In this work, we introduce EvoPER, an R package for simplifying the parameter estimation using evolutionary computation methods.

Alfonso Rodríguez-Patón

Papers

Towards a robust biocomputing solution of intractable problems

On the power of families of recognizer spiking neural p systems

Probabilistic Reasoning with an Enzyme-Driven DNA Device

Desarrollo de Aplicaciones de la Computación Celular con Membranas y la Computación Biomolecular en la Biología TIN2006-15595

Applying Evolutionary Metaheuristics for Parameter Estimation of Individual-Based Models