A Hierarchical Clustering and Fixed-Layer Local Learning Based Support Vector Machine Algorithm for Large Scale Classification Problems

It is a challenging topic to develop an efficient algorithm for large scale classification problems in many applications of machine learning. In this paper, a hierarchical clustering and fixed- layer local learning （HCFLL） based support vector machine（SVM） algorithm is proposed to deal with this problem. Firstly, HCFLL hierarchically dusters a given dataset into a modified clustering feature tree based on the ideas of unsupervised clustering and supervised clustering. Then it locally trains SVM on each labeled subtree at a fixed-layer of the tree. The experimental results show that compared with the existing popular algorithms such as core vector machine and decision.tree support vector machine, HCFLL can significantly improve the training and testing speeds with comparable testing accuracy.

Prognostic factors and its predictive value in patients with metastatic spinal cancer

BACKGROUND The spine is the most common location of metastatic diseases.Treating a metastatic spinal tumor depends on many factors,including patients’overall health and life expectancy.The present study was conducted to investigate prognostic factors and clinical outcomes in patients with vertebral metastases.AIM To investigate prognostic factors and their predictive value in patients with metastatic spinal cancer.METHODS A retrospective analysis of 109 patients with metastatic spinal cancer was conducted between January 2015 and September 2017.The prognoses and survival were analyzed,and the effects of factors such as clinical features,treatment methods,primary lesions and affected spinal segments on the prognosis of patients with metastatic spinal cancer were discussed.The prognostic value of Frankel spinal cord injury functional classification scale,metastatic spinal cord compression(MSCC),spinal instability neoplastic score(SINS)and the revised Tokuhashi score for prediction of prognosis was explored in patients with metastatic spinal tumors.RESULTS Age,comorbidity of metastasis from elsewhere,treatment methods,the number of spinal tumors,patient’s attitude toward tumors and Karnofsky performance scale score have an effect on the prognosis of patients(all P<0.05).With respect to classification of spinal cord injury,before operation,the proportion of grade B and grade C was higher in the group of patients who died than in the group of patients who survived,and that of grade D and grade E was lower in the group of patients who died than in the group of patients who survived(all P<0.05).At 1 mo after operation,the proportion of grade A,B and C was higher in the group of patients who died than in the group of patients who survived,and that of grade E was lower in patients in the group of patients who died than in the group of patients who survived(all P<0.05).MSCC occurred in four(14.3%)patients in the survival group and 17(21.0%)patients in the death group(P<0.05).All patients suffered from intractable pain,dysfunction in spinal cord and even paralysis.The proportion of SINS score of 1 to 6 points was lower in the death group than in the survival group,and the proportion of SINS score of 7 to 12 points was higher in the death group than in the survival group(all P<0.05).The proportion of revised Tokuhashi score of 0 to 8 points and 9 to 11 points were higher in the death group than in the survival group,and the proportion of revised Tokuhashi score of 12 to 15 points was lower in the death group than in the survival group(all P<0.05).Frankel spinal cord injury functional classification scale,MSCC,SINS and revised Tokuhashi score were important factors influencing the surgical treatment of patients with metastatic spinal cancer(all P<0.05).CONCLUSION Frankel spinal cord injury functional classification scale,MSCC,SINS and revised Tokuhashi score were helpful in predicting the prognosis of patients with metastatic spinal cancer.

Dependence of low redshift Type Ia Supernovae luminosities on host galaxies

We study the relation between Type Ia Supernovae （SNe Ia） and properties of their host galaxies using a large sample with low redshift. By examining the Hubble residuals of the entire sample from the best-fit cosmology, we show that SNe Ia in passive hosts are brighter than those in star-forming hosts after light curve correction at the 2. 1σ confidence level. We find that SNe Ia in high luminosity hosts are brighter after light-curve correction at the 〉 3σ confidence level. We also find that SNe Ia in large galaxies are brighter after light-curve correction at the ≥2σ confidence level. We demonstrate that the residuals depend linearly on host luminosity at a confidence of 4or or host size at a confidence of 3.3σ.

Image categorization with resource constraints： introduction, challenges and advances

As one of the most classic fields in computer vi- sion, image categorization has attracted widespread interests. Numerous algorithms have been proposed in the community, and many of them have advanced the state-of-the-art. How- ever, most existing algorithms are designed without consider- ation for the supply of computing resources. Therefore, when dealing with resource constrained tasks, these algorithms will fail to give satisfactory results. In this paper, we provide a comprehensive and in-depth introduction of recent develop- ments of the research in image categorization with resource constraints. While a large portion is based on our own work, we will also give a brief description of other elegant algo- rithms. Furthermore, we make an investigation into the re- cent developments of deep neural networks, with a focus on resource constrained deep nets.