We present a method for learning sparse representations shared across multiple tasks. This method is a generalization of the well-known single-task 1-norm regularization. It is based on a novel non-convex regularizer which controls the number of learned features common across the tasks. We prove that the method is equivalent to solving a convex optimization problem for which there is an iterative algorithm which converges to an optimal solution. The algorithm has a simple interpretation: it alternately performs a supervised and an unsupervised step, where in the former step it learns task-specific functions and in the latter step it learns common-across-tasks sparse representations for these functions. We also provide an extension of the algorithm which learns sparse nonlinear representations using kernels. We report experiments on simulated and real data sets which demonstrate that the proposed method can both improve the performance relative to learning each task independently and lead to a few learned features common across related tasks. Our algorithm can also be used, as a special case, to simply select-not learn-a few common variables across the tasks.
Summary. We present a method for learning sparse representations shared across multiple tasks. This method is a generalization of the well-known singletask 1-norm regularization. It is based on a novel non-convex regularizer which controls the number of learned features common across the tasks. We prove that the method is equivalent to solving a convex optimization problem for which there is an iterative algorithm which converges to an optimal solution. The algorithm has a simple interpretation: it alternately performs a supervised and an unsupervised step, where in the former step it learns task-specific functions and in the latter step it learns common-across-tasks sparse representations for these functions. We also provide an extension of the algorithm which learns sparse nonlinear representations using kernels. We report experiments on simulated and real data sets which demonstrate that the proposed method can both improve the performance relative to learning each task independently and lead to a few learned features common across related tasks. Our algorithm can also be used, as a special case, to simply select -not learn -a few common variables across the tasks 3 .
Commonly used chemotherapeutic agents in oncology/hematology practice, causing toxic peripheral neuropathy, include taxanes, platinum compounds, vinca alkaloids, proteasome inhibitors, and antiangiogenic/immunomodulatory agents. This review paper intends to put together and discuss the spectrum of chemotherapy-induced peripheral neuropathy (CIPN) characteristics so as to highlight areas of future research to pursue on the topic. Current knowledge shows that the pathogenesis of CIPN still remains elusive, mostly because there are several sites of involvement in the peripheral nervous system. In any case, it is acknowledged that the dorsal root ganglia of the primary sensory neurons are the most common neural targets of CIPN. Both the incidence and severity of CIPN are clinically under- and misreported, and it has been demonstrated that scoring CIPN with common toxicity scales is associated with significant inter-observer variability. Only a proportion of chemotherapy-treated patients develop treatment-emergent and persistent CIPN, and to date it has been impossible to predict high-and low-risk subjects even within groups who receive the same drug regimen. This issue has recently been investigated in the context of pharmacogenetic analyses, but these studies have not implemented a proper methodological approach and their results are inconsistent and not really clinically relevant. As such, a stringent approach has to be implemented to validate that information. Another open issue is that, at present, there is insufficient evidence to support the use of any of the already tested chemoprotective agents to prevent or limit CIPN. The results of comprehensive interventions, including clinical, neurophysiological, and pharmacogenetic approaches, are expected to produce a consistent advantage for both doctors and patients and thus allow the registration and analysis of reliable data on the true characteristics of CIPN, eventually leading to potential preventive and therapeutic interventions.
Background: Chemotherapy-induced peripheral neuropathy (CIPN) is a debilitating and dose-limiting complication of cancer treatment. Thus far, the impact of CIPN has not been studied in a systematic clinimetric manner. The objective of the study was to select outcome measures for CIPN evaluation and to establish their validity and reproducibility in a cross-sectional multicenter study.
Patients and methods:After literature review and a consensus meeting among experts, face/content validity were obtained for the following selected scales: the National Cancer Institute-Common Toxicity Criteria (NCI-CTC), the Total Neuropathy Score clinical version (TNSc), the modified Inflammatory Neuropathy Cause and Treatment (INCAT) group sensory sumscore (mISS), the European Organization for Research and Treatment of Cancer (EORTC) QLQ-C30, and CIPN20 quality-of-life measures. A total of 281 patients with stable CIPN were examined. Validity (correlation) and reliability studies were carried out.Results: Good inter-/intra-observer scores were obtained for the TNSc, mISS, and NCI-CTC sensory/motor subscales. Test-retest values were also good for the EORTC QLQ-C30 and CIPN20. Acceptable validity scores were obtained through the correlation among the measures.
Conclusion:Good validity and reliability scores were demonstrated for the set of selected impairment and quality-of-life outcome measures in CIPN. Future studies are planned to investigate the responsiveness aspects of these measures.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.