The fungicidal activities of Cassia tora extracts and their active principles were determined against Botrytis cineria, Erysiphe graminis, Phytophthora infestans, Puccinia recondita, Pyricularia grisea, and Rhizoctonia solani using a whole plant method in vivo and were compared with synthetic fungicides and three commercially available anthraquinones. The responses varied with the plant pathogen tested. At 1 g/L, the chloroform fraction of C. tora showed a strong fungicidal activity against B. cinerea, E. graminis, P. infestans, and R. solani. Emodin, physcion, and rhein were isolated from the chloroform fraction using chromatographic techniques and showed strong and moderate fungicidal activities against B. cinerea, E. graminis, P. infestans, and R. solani. Furthermore, aloe-emodin showed strong and moderate fungicidal activities against B. cinerea and R. solani, respectively, but did not inhibit the growth of E. graminis, P. infestans, P. recondita, and Py. grisea. Little or no activity was observed for anthraquinone and anthraquinone-2-carboxylic acid when tested at 1 g/L. Chlorothalonil and dichlofluanid as synthetic fungicides were active against P. infestans and B. cinerea at 0.05 g/L, respectively. Our results demonstrate the fungicidal actions of emodin, physcion, and rhein from C. tora.

We present a new semi-supervised training procedure for conditional random fields (CRFs) that can be used to train sequence segmentors and labelers from a combination of labeled and unlabeled training data. Our approach is based on extending the minimum entropy regularization framework to the structured prediction case, yielding a training objective that combines unlabeled conditional entropy with labeled conditional likelihood. Although the training objective is no longer concave, it can still be used to improve an initial model (e.g. obtained from supervised training) by iterative ascent. We apply our new training algorithm to the problem of identifying gene and protein mentions in biological texts, and show that incorporating unlabeled data improves the performance of the supervised CRF in this case.

The memory reconsolidation hypothesis suggests that a memory trace becomes labile after retrieval and needs to be reconsolidated before it can be stabilized. However, it is unclear from earlier studies whether the same synapses involved in encoding the memory trace are those that are destabilized and restabilized after the synaptic reactivation that accompanies memory retrieval, or whether new and different synapses are recruited. To address this issue, we studied a simple nonassociative form of memory, long-term sensitization of the gill-and siphon-withdrawal reflex in Aplysia, and its cellular analog, long-term facilitation at the sensory-to-motor neuron synapse. We found that after memory retrieval, behavioral long-term sensitization in Aplysia becomes labile via ubiquitin/proteasome-dependent protein degradation and is reconsolidated by means of de novo protein synthesis. In parallel, we found that on the cellular level, longterm facilitation at the sensory-to-motor neuron synapse that mediates long-term sensitization is also destabilized by protein degradation and is restabilized by protein synthesis after synaptic reactivation, a procedure that parallels memory retrieval or retraining evident on the behavioral level. These results provide direct evidence that the same synapses that store the long-term memory trace encoded by changes in the strength of synaptic connections critical for sensitization are disrupted and reconstructed after signal retrieval. memory reorganization | memory recall | 5-HT | local protein synthesis | clasto-lactacystin beta-lactone T he processes of memory reactivation (retrieval) have been the focus of several studies over the last decade. Retrieval is thought to return the memory to an unstable (labile) state, in which de novo protein synthesis-dependent reconsolidation is required to continue maintaining the memory over time (1-4). Memory reconsolidation has been reported for a variety of memory paradigms in a number of different animal models (1, 3, 5, 6); however, how memory reconsolidation works remains unclear.At least two nonmutually exclusive hypotheses have been proposed (7). One hypothesis suggests that reconsolidation is an updating process in which the synapses that encode the preexisting memory are reorganized after memory retrieval so as to recruit new synaptic connections that allow the incorporation of new information (8-10). The second hypothesis suggests a mechanism that is a continuation of the consolidation process at the same set of synaptic connections and that serves to strengthen memory, allowing it to become longer lasting and enduring and thereby preventing forgetting (11). Both of these views of reconsolidation are consistent with retraining or retrieval. In each case, synaptic reactivation could be implicit (e.g., during sleep) or explicit, and both would presumably have the same effect of making the memory stronger, more stable, and more resistant to postretrieval interference.Both types of reconsolidation hypotheses imply that the stored memory become...

No abstract

Abstract. In this paper we propose Support Vector Random Fields (SVRFs), an extension of Support Vector Machines (SVMs) that explicitly models spatial correlations in multi-dimensional data. SVRFs are derived as Conditional Random Fields that take advantage of the generalization properties of SVMs. We also propose improvements to computing posterior probability distributions from SVMs, and present a local-consistency potential measure that encourages spatial continuity. SVRFs can be efficiently trained, converge quickly during inference, and can be trivially augmented with kernel functions. SVRFs are more robust to class imbalance than Discriminative Random Fields (DRFs), and are more accurate near edges. Our results on synthetic data and a real-world tumor detection task show the superiority of SVRFs over both SVMs and DRFs.

No abstract

No abstract

Abstract. Locating Brain tumor segmentation within MR (magnetic resonance) images is integral to the treatment of brain cancer. This segmentation task requires classifying each voxel as either tumor or nontumor, based on a description of that voxel. Unfortunately, standard classifiers, such as Logistic Regression (LR) and Support Vector Machines (SVM), typically have limited accuracy as they treat voxels as independent and identically distributed (iid ). Approaches based on random fields, which are able to incorporate spatial constraints, have recently been applied to brain tumor segmentation with notable performance improvement over iid classifiers. However, previous random field systems involved computationally intractable formulations, which are typically solved using some approximation. Here, we present pseudo-conditional random fields (PCRFs), which achieve accuracy similar to other random fields variants, but are significantly more efficient. We formulate a PCRF as a regularized discriminative classifier that relaxes the classification decision for each voxel by considering the labels and features of neighboring voxels.