We present a general method using kernel canonical correlation analysis to learn semantic representation web images and their associated text. The space provides common enables comparison between the text images. In experiments, we look at two approaches of retrieving based on only content from query. compare orthogonalization against standard cross-representation retrieval technique known as generalized vector model.

Objective This study proposes to assess the differences of two psychosocial risk indicators for coronary artery disease (CAD), ie, depressive symptoms and vital exhaustion. Method In a representative, stratified, nation-wide sample population Hungary over age 16 years (N = 12,640), analyses were made whether those differentially related several illness behaviors (including history cardiovascular treatment sick days), cognitions, mood states, socioeconomic characteristics that may generally...

Abstract We present comboFM, a machine learning framework for predicting the responses of drug combinations in pre-clinical studies, such as those based on cell lines or patient-derived cells. comboFM models context-specific interactions through higher-order tensors, and efficiently learns latent factors tensor using powerful factorization machines. The approach enables to leverage information from previous experiments performed similar drugs cells when new so far untested cells; thereby, it...

Abstract Prediction of drug combination responses is a research question growing importance for cancer and other complex diseases. Current machine learning approaches generally consider predicting either synergy summaries or single dose-response values, which fail to appropriately model the continuous nature underlying surface can lead inconsistencies when score matrix reconstructed from separate predictions. We propose novel prediction method, comboKR, that directly predicts response...

Many inference problems in bioinformatics, including drug bioactivity prediction, can be formulated as pairwise learning problems, which one is interested making predictions for pairs of objects, e.g. drugs and their targets. Kernel-based approaches have emerged powerful tools solving that kind, especially multiple kernel (MKL) offers promising benefits it enables integrating various types complex biomedical information sources the form kernels, along with importance prediction task....

Abstract Motivation Liquid Chromatography (LC) followed by tandem Mass Spectrometry (MS/MS) is one of the predominant methods for metabolite identification. In recent years, machine learning has started to transform analysis mass spectra and identification small molecules. contrast, LC data rarely used improve identification, despite numerous published retention time prediction using learning. Results We present a method predicting order molecules; that is, in which molecules elute from...

Abstract Background In last two decades, the use of high-throughput sequencing technologies has accelerated pace discovery proteins. However, due to time and resource limitations rigorous experimental functional characterization, functions a vast majority them remain unknown. As result, computational methods offering accurate, fast large-scale assignment new previously unannotated proteins are sought after. Leveraging underlying associations between multiplicity features that describe could...

We present a kernel-based algorithm for hierarchical text classification where the documents are allowed to belong more than one category at time. The model is variant of Maximum Margin Markov Network framework, hierarchy represented as tree equipped with an exponential family defined on edges. efficient optimization based incremental conditional gradient ascent in single-example subspaces spanned by marginal dual variables. Experiments show that can feasibly optimize training sets thousands...

Engineered microbial cells present a sustainable alternative to fossil-based synthesis of chemicals and fuels. Cellular routes are readily assembled introduced into strains using state-of-the-art synthetic biology tools. However, the optimization required reach industrially feasible production levels is far less efficient. It typically relies on trial-and-error leading high uncertainty in total duration cost. New techniques that can cope with complexity limited mechanistic knowledge cellular...

Abstract Background In this paper we describe work in progress developing kernel methods for enzyme function prediction. Our focus is so called structured output prediction methods, where the enzymatic reaction combinatorial target object We compared two Hierarchical Max-Margin Markov algorithm (HM 3 ) and Maximum Margin Regression (MMR) hierarchical classification of function. As sequence features use various string kernels GTG feature set derived from global alignment trace graph protein...

Abstract Drug combinations are required to treat advanced cancers and other complex diseases. Compared with monotherapy, combination treatments can enhance efficacy reduce toxicity by lowering the doses of single drugs—and there especially synergistic interest. Since drug screening experiments costly time-consuming, reliable machine learning models needed for prioritizing potential further studies. Most current based on scalar-valued approaches, which predict individual response values or...

In complex manipulation scenarios (e.g. tasks requiring interaction of two hands or in-hand manipulation), generalization is a hard problem. Current methods still either require substantial amount (supervised) training data and / strong assumptions on both the environment task. this paradigm, controllers solving these tend to be complex. We propose paradigm maintaining simpler task in small number specific situations. order generalize novel situations, robot transforms from situations into...

Abstract Motivation Combination therapies have emerged as a powerful treatment modality to overcome drug resistance and improve efficacy. However, the number of possible combinations increases very rapidly with individual drugs in consideration, which makes comprehensive experimental screening infeasible practice. Machine-learning models offer time- cost-efficient means aid this process by prioritizing most effective for further pre-clinical clinical validation. complexity underlying...

The aim of this paper is to propose a system where complex affordance learning bootstrapped through using pre-learned basic-affordances as additional inputs the predictors or cues in selecting next objects explore during learning. In first stage, robot learns affordances form developing classifiers that predict effect categories given object features for different discrete actions applicable single objects. These predictions are later added robot's feature set higher-level features. second...