Leveraging Transfer Learning for Spatio-Temporal Human Activity Recognition from Video Sequences

Human Activity Recognition (HAR) is an active research area due to its applications in pervasive computing, human-computer interaction, artificial intelligence, health care, and social sciences. Moreover, dynamic environments and anthropometric differences between individuals make it harder to recognize actions. This study focused on human activity in video sequences acquired with an RGB camera because of its vast range of real-world applications. It uses two-stream ConvNet to extract spatial and temporal information and proposes a fine-tuned deep neural network. Moreover, the transfer learning paradigm is adopted to extract varied and fixed frames while reusing object identification information. Six state-of-the-art pre-trained models are exploited to find the best model for spatial feature extraction. For temporal sequence, this study uses dense optical flow following the two-stream ConvNet and Bidirectional Long Short Term Memory (BiLSTM) to capture long-term dependencies. Two state-of-the-art datasets, UCF101 and HMDB51, are used for evaluation purposes. In addition, seven state-of-the-art optimizers are used to fine-tune the proposed network parameters. Furthermore, this study utilizes an ensemble mechanism to aggregate spatial-temporal features using a four-stream Convolutional Neural Network (CNN), where two streams use RGB data. In contrast, the other uses optical flow images. Finally, the proposed ensemble approach using max hard voting outperforms state-of-the-art methods with 96.30% and 90.07% accuracies on the UCF101 and HMDB51 datasets.     

Matching preparatory intervention to coping style: The effects on children's distress in the dental setting

OBJECTIVE: Investigate the effects of matching preparatory interventions to patient's coping styles. METHODS: Participants were 61 children, with a restricted age range of 6 through 9 years old (mean age was 7.9 years), who underwent dental restoration. Participants were randomly assigned to an information intervention, a relaxation intervention, or a control condition. Play and parent-report of sensitization/repression were indices of coping style. The first hypothesis, that play would relate to sensitization/repression, was tested using Pearson correlations. The second hypothesis, that interventions that were congruent with patients' coping styles would be more effective than incongruent interventions, was tested using MANCOVAs. RESULTS: No relation was found between play and coping style. The "congruency hypothesis" was supported for self-reported distress immediately following the intervention. On behavioral distress variables, the interaction between sensitization/repression and condition was contrary to the congruency hypothesis. CONCLUSIONS: Implications for future research and clinical intervention with pediatric populations were discussed.     

CD57+/CD28- T cells in untreated hemato-oncological patients are expanded and display a Th1-type cytokine secretion profile, ex vivo cytolytic activity and enhanced tendency to apoptosis

Protopolystoma (Monogenea, Polystomatidae) is strictly specific to the anuran amphibian genus Xenopus. The host group is characterised by a polyploid series in which chromosome numbers reflect diploid, tetraploid, octoploid and dodecaploid constitutions; the series is considered to have evolved through interspecies hybridisation and genome duplication. This study correlates information on host evolutionary relationships with patterns of parasite speciation and host specificity. Protopolystoma is restricted to one subgenus (Xenopus) with multiples of 36 chromosomes, and is absent from the subgenus Silurana (with multiples of 20 chromosomes). Molecular, biochemical and karyotype evidence distinguishes three subgroups within Xenopus. Representative species from each subgroup, Xenopus muelleri, Xenopus fraseri and Xenopus laevis, have been examined for polystomatid infection. Two species of Protopolystoma occur in each of these host species. In X. muelleri, the two Protopolystoma species reflect parasite co-speciation corresponding with the divergence of two sibling host species. Xenopus fraseri and X. laevis (both with 2n = 36 chromosomes) are implicated in the hybrid origin of two octoploid species, Xenopus wittei and Xenopus vestitus (both 2n = 72). The relationships of the Protopolystoma species in these Xenopus taxa reflect this presumed ancestry. Xenopus wittei carries two species of Protopolystoma, one shared with X. fraseri and the other shared with X. laevis. Xenopus vestitus carries a single species of Protopolystoma which is shared with X. laevis but there is no "heirloom" which reflects its hybrid origin involving X. fraseri. In addition to these shared parasite species which may reflect shared host genes, X. fraseri and X. laevis each carry separate species-specific Protopolystoma which do not occur in other Xenopus species even where there is evidence of common genetic information (as in the allopolyploid wittei and vestitus). This case study may be interpreted as indicating a powerful influence of host genetic factors on susceptibility to infection, host-specificity, and parasite speciation.     

Surgical anatomy for endoscopic subfascial division of perforating veins

PURPOSE: This study was undertaken to define the surgical anatomy of the medial perforating veins (PVs) of the leg and to provide information on how to gain access to all medial PVs from the superficial posterior compartment during a subfascial endoscopic procedure. METHODS: The venous anatomy of 40 limbs (from 23 cadavers) were studied. Medial PVs located between the ankle and the tibial tuberosity were dissected. None of the subjects had pathologic evidence of venous disease. Each PV's type (direct or indirect), size (< 1 mm, 1 to 2 mm, > 2 mm), location (distances from ankle [D1], and tibia [D2]), and accessibility from the superficial posterior compartment were recorded. RESULTS: Five hundred fifty-two PVs were identified (mean, 13.8; range, 7 to 22). Two hundred eighty-seven PVs (52%) directly connected the superficial with the deep systems, 228 (41%) were indirect muscle perforators, and 37 PVs (7%) were undetermined. One hundred thirty-seven PVs (25%) were > 2 mm. Sixty-three percent of PVs were accessible from the superficial posterior compartment. In the distal half of the leg, two groups of direct PVs could be identified (Cockett II: D1, 7 to 9 cm; Cockett III: D1, 10 to 12 cm). In the proximal half of the leg, paratibial direct PVs (D2 < or = 1 cm) were found clustered in three groups (D1, 18 to 22 cm; D1, 23 to 27 cm; D1, 28 to 32 cm). CONCLUSIONS: Our study confirmed the presence of the Cockett II and III PVs and three groups of proximal paratibial PVs, including the "24-cm" perforators. Two thirds of the medial direct PVs are accessible for endoscopic division from the superficial posterior compartment. To divide paratibial PVs, however, incision of the paratibial deep fascia is frequently required.     

Iterative optimization of high-affinity proteases inhibitors using phage display. 1. Plasmin

We generated a series of libraries having variants of the first Kunitz domain of human lipoprotein-associated coagulation inhibitor (LACI-D1, also known as tissue-factor pathway inhibitor-I) displayed on bacteriophage M13 as pIII-fusions. We varied LACI-DI iteratively in two regions: the P1 region (positions 10-21) and the "second loop", (positions 31-39), which together form one end of the domain. Display-phage library Lib#1 allows 31 200 amino-acid sequences in P1 region (residues 13, 16-19). Preliminary, we screened Lib#1 against human plasmin (PLA, EC 3.4.21.7) immobolized on agarose to enrich for phage displaying variants with PLA affinity. We introduced a 1600-fold increase in second-loop diversity (residues 31, 32, 34, 39) into the population of selectants from Lib#1, yielding Lib#2. Lib#2 (allowing approximately 50 million amino-acid sequences) was screened against PLA-agarose to isolate highest affinity binders. Protein EPI-P211, derived from the best isolate of Lib#2, inhibits PLA with Ki = 2 nM (at least 500-fold better than LACI-D1) and with high specificity. We used amino-acid sequences of PLA-binding selectants to design a PLA-biased library (Lib#3) which we screened against PLA. The protein EPI-P302 (derived from the best binder obtained from Lib#3) has Ki for PLA inhibition of 87 pM, which is 25-fold better than the first-round best binder and > or = 12 500-fold better than LACI-D1. EPI-P302 also shows very high specificity for PLA vs other human proteases and is resistant to inactivation by oxidants and extremes of temperature or pH. Thus, one can use selectants from one library to design target-tailored combinatorial libraries and obtain quite stable, highly specific, very high-affinity binding molecules while maintaining an essentially human framework.     

Perinuclear antineutrophil cytoplasmic antibodies in patients with Crohn's disease define a clinical subgroup

BACKGROUND & AIMS: Antineutrophil cytoplasmic antibodies (ANCA) have been consistently detected in a subgroup of patients with Crohn's disease (CD). This study was designed to determine whether serum ANCA expression in patients with CD characterizes an identifiable clinical subgroup. METHODS: The study population consisted of 69 consecutive patients with an established diagnosis of CD as determined by a combination of characteristic clinical, radiographic, endoscopic, and histopathologic criteria. Sera from the patients were analyzed for the presence of ANCAs using the fixed neutrophil enzyme-linked immunosorbent assay (ELISA) assay. Perinuclear ANCA (pANCA)-positive and cytoplasmic ANCA (cANCA)-positive results by ELISA were confirmed by indirect immunofluorescence staining. Clinical profiles of the ANCA-positive patients with CD were compared with those of patients with CD not expressing ANCA (ANCA-negative). RESULTS: pANCA-positive patients with CD have endoscopically and/or histopathologically documented left-sided colitis and symptoms of left-sided colonic inflammation, clinically reflected by rectal bleeding and mucus discharge, urgency, and treatment with topical agents. One hundred percent of patients with CD expressing pANCA had "UC-like" features. CONCLUSIONS: In patients with CD, serum pANCA expression characterizes a UC-like clinical phenotype. Stratification of CD by serum pANCA provides evidence of heterogeneity within CD and suggests a common intestinal mucosal inflammatory process among a definable subgroup of patients with CD and UC expressing this marker.     

ManiWordle: providing flexible control over Wordle

Among the multifarious tag-clouding techniques, Wordle stands out to the community by providing an aesthetic layout, eliciting the emergence of the participatory culture and usage of tag-clouding in the artistic creations. In this paper, we introduce ManiWordle, a Wordle-based visualization tool that revamps interactions with the layout by supporting custom manipulations. ManiWordle allows people to manipulate typography, color, and composition not only for the layout as a whole, but also for the individual words, enabling them to have better control over the layout result. We first describe our design rationale along with the interaction techniques for tweaking the layout. We then present the results both from the preliminary usability study and from the comparative study between ManiWordle and Wordle. The results suggest that ManiWordle provides higher user satisfaction and an efficient method of creating the desired "art work," harnessing the power behind the ever-increasing popularity of Wordle.     

Underdetermined Convolutive Blind Source Separation via Time–Frequency Masking

In this paper, we consider the problem of separation of unknown number of sources from their underdetermined convolutive mixtures via time-frequency (TF) masking. We propose two algorithms, one for the estimation of the masks which are to be applied to the mixture in the TF domain for the separation of signals in the frequency domain, and the other for solving the permutation problem. The algorithm for mask estimation is based on the concept of angles in complex vector space. Unlike the previously reported methods, the algorithm does not require any estimation of the mixing matrix or the source positions for mask estimation. The algorithm clusters the mixture samples in the TF domain based on the Hermitian angle between the sample vector and a reference vector using the well known k -means or fuzzy c -means clustering algorithms. The membership functions so obtained from the clustering algorithms are directly used as the masks. The algorithm for solving the permutation problem clusters the estimated masks by using k-means clustering of small groups of nearby masks with overlap. The effectiveness of the algorithm in separating the sources, including collinear sources, from their underdetermined convolutive mixtures obtained in a real room environment, is demonstrated.