Synthetic Data in Human Analysis: A Survey.

Deep neural networks have become prevalent in human analysis, boosting the performance of applications, such as biometric recognition, action recognition, as well as person re-identification. However, the performance of such networks scales with the available training data. In human analysis, the demand for large-scale datasets poses a severe challenge, as data collection is tedious, time-expensive, costly and must comply with data protection laws. Current research investigates the generation of synthetic data as an efficient and privacy-ensuring alternative to collecting real data in the field. This survey introduces the basic definitions and methodologies, essential when generating and employing synthetic data for human analysis. We summarise current state-of-the-art methods and the main benefits of using synthetic data. We also provide an overview of publicly available synthetic datasets and generation models. Finally, we discuss limitations, as well as open research problems in this field. This survey is intended for researchers and practitioners in the field of human analysis.

Clustering Single-Cell RNA Sequence Data Using Information Maximized and Noise-Invariant Representations.

Single-cell RNA sequencing (scRNA-seq) is a revolutionary methodology that helps to analyze transcriptome or genome information from a single cell. However, high dimensionality and sparsity in data due to dropout events pose computational challenges for existing state-of-the-art scRNA-seq clustering methods. Learning efficient representations becomes even more challenging due to the presence of noise in scRNA-seq data. To overcome the effect of noise and learn effective representations, this paper proposes sc-INDC (Single-Cell Information Maximized Noise-Invariant Deep Clustering), a deep neural network that facilitates learning of informative and noise-invariant representations of scRNA-seq data. Furthermore, the time complexity of the proposed sc-INDC is significantly lower compared to state-of-the-art scRNA-seq clustering methods. Extensive experimentation on fourteen publicly available scRNA-seq datasets illustrates the efficacy of the proposed model. Additionally, visualizations of t-SNE plots and several ablation studies are also conducted to provide insights into the improved representation ability of sc-INDC. Code of the proposed sc-INDC will be available at: https://github.com/arnabkmondal/sc-INDC.

Leveraging joint incremental learning objective with data ensemble for class incremental learning.

A class-incremental learning problem is characterized by training data becoming available in a phase-by-phase manner. Deep learning models suffer from catastrophic forgetting of the classes in the older phases as they get trained on the classes introduced in the new phase. In this work, we show that the change in orientation of an image has a considerable effect on the model prediction accuracy, which in turn demonstrates the different rates of catastrophic forgetting for the different orientations of the same image, which is a novel finding. Based on this, we propose a data-ensemble approach that combines the predictions for the different orientations of the image to help the model retain information regarding the previously seen classes and thereby reduce the rate of forgetting in the model predictions. However, we cannot directly use the data-ensemble approach if the model is trained using traditional techniques. Therefore, we also propose a novel training approach using a joint-incremental learning objective (JILO) that involves jointly training the network with two incremental learning objectives, i.e., the class-incremental learning objective and our proposed data-incremental learning objective. We empirically demonstrate that JILO is vital to the data-ensemble approach. We apply our proposed approach to state-of-the-art class-incremental learning methods and empirically show that our approach significantly improves the performance of these methods. Our proposed approach significantly improves the performance of the state-of-the-art method (AANets) on the CIFAR-100 dataset by absolute margins of 3.30%, 4.28%, 3.55%, 4.03%, for the number of phases P=50, 25, 10, and 5, respectively, which establishes the efficacy of the proposed work.

Coincident activation of metabotropic glutamate receptors and NMDA receptors (NMDARs) downregulates perisynaptic/extrasynaptic NMDARs and enhances high-fidelity neurotransmission at the developing calyx of Held synapse.

NMDA receptors (NMDARs) are usually downregulated in developing central synapses, but underlying mechanisms and functional consequences are not well established. Using developing calyx of Held synapses in the mouse auditory brainstem, we show here that pairing presynaptic stimulation with postsynaptic depolarization results in a persistent downregulation in the summated amplitude of NMDAR-mediated EPSCs (NMDAR-EPSCs) during a train of stimuli (100/200 Hz, 100 ms) at both 22 degrees C and 35 degrees C. In contrast, the amplitude of single NMDAR-EPSCs or AMPA receptor-mediated EPSCs in the same synapses is not significantly altered, implying a preferential downregulation of perisynaptic/extrasynaptic NMDARs. Induction of this downregulation is blocked by antagonists for NMDARs or group I metabotropic glutamate receptors (mGluRs), suggesting that coincident activation of these two receptors is required. When the postsynaptic neuron is loaded with the fast Ca2+ buffer BAPTA or depolarized to +60 mV to reduce the driving force for Ca2+ influx, downregulation of the summated NMDAR-EPSCs is abolished, indicating Ca2+ plays a critical role in the induction. The expression of this downregulation depends on ongoing synaptic activity, and is attenuated by a dynamin peptide (D15) that blocks clathrin-dependent internalization. We further demonstrated that the same induction paradigm specifically reduces NMDAR-dependent plateau potential and aberrant spike firings during repetitive activity. Together, our results suggest that coincident activation of mGluRs and NMDARs during intense synaptic activity may lead to selective endocytosis of NMDARs in the perisynaptic/extrasynaptic domain, and implicate that mGluRs are potentially important for gating development of high-fidelity neurotransmission at this synapse.

Pregabalin action at a model synapse: binding to presynaptic calcium channel alpha2-delta subunit reduces neurotransmission in mice.

Pregabalin, ((S)-3-(aminomethyl)-5-methylhexanoic acid, also known as (S)-3-isobutyl GABA, Lyricatrade mark) is approved for treatment of certain types of peripheral neuropathic pain and as an adjunctive therapy for partial seizures of epilepsy both the EU and the USA and also for generalized anxiety disorder in the EU. Though pregabalin binds selectively to the alpha(2)-delta (alpha(2)-delta) auxiliary subunit of voltage-gated calcium channels, the cellular details of pregabalin action are unclear. The high density of alpha(2)-delta in skeletal muscle fibers raises the question of whether pregabalin alters excitation-contraction coupling. We used the mouse soleus neuromuscular junction from mice containing an artificially mutated alpha(2)-delta Type 1 protein (R217A) as a model to examine the effect of pregabalin. Pregabalin reduced nerve-evoked muscle contractions by 16% at a clinically relevant concentration of 10 muM in wildtype mice. When acetylcholine receptors were blocked with curare, pregabalin had no effect on contraction from direct stimulation of muscle, suggesting a lack of drug effects on contraction coupling. Our data are consistent with pregabalin having no effect on striated muscle L-type calcium channel function. However, in mice expressing mutant (R217A) alpha(2)-delta Type 1, there was no significant effect of pregabalin on nerve-evoked muscle contraction. We propose that pregabalin reduces presynaptic neurotransmitter release without altering postsynaptic receptors or contraction coupling and that these effects require high affinity binding to alpha(2)-delta Type 1 auxiliary subunit of presynaptic voltage-gated calcium channels.

The role of AMPA receptor gating in the development of high-fidelity neurotransmission at the calyx of Held synapse.

During early postnatal development of auditory synapses, the decay time course of AMPA receptor (AMPAR) EPSCs accelerates markedly, but the mechanisms underlying this process remain uncertain. Using the developing calyx of Held synapse in the mouse auditory brainstem, we have examined presynaptic and postsynaptic elements that may regulate decay kinetics of AMPAR EPSCs. We found that the decay time kinetics was voltage dependent in both immature and mature synapses, being slower at positive potentials than negative potentials. By recording evoked miniature events in extracellular Ca2+ or Sr2+, we revealed a significant decrease in decay time constants of EPSCs as maturation progresses. On the basis of internal and external polyamine block of AMPAR EPSCs and immunohistochemistry assays with subunit-specific antibodies, we demonstrated that the glutamate receptor (GluR) 2 subunit is virtually absent at all developmental ages. Antibody staining patterns suggest a gradual shift in subunit composition from GluR1- to GluR3/4-dominant phenotypes. Kinetic analyses of deactivation, desensitization, and recovery from desensitization in outside-out patches in response to ultrafast application of glutamate lend supportive evidence that such a shift in the gating phenotype likely accounts for the accelerated time course throughout development. Finally, by pharmacologically manipulating AMPAR gating and using simulated EPSCs to evoke action potentials, we demonstrated that rapid decay kinetics of AMPAR EPSCs is essential for this synapse to accommodate high-frequency firing without compromising spike amplitude. Hence, developmental alterations in the subunit composition likely dictate changes in the time course of AMPAR EPSCs and play an indispensable role in the refinement of high-fidelity neurotransmission at the calyx of Held synapse.

A psychometric cut-point to separate emergently suicidal depressed patients from stable depressed outpatients.

CONTEXT: The design of safe clinical trials targeting suicidal ideation requires operational definitions of what degree of suicidal ideation is too excessive to allow safe participation. AIMS: We examined the Scale for Suicide Ideation (SSI) to develop a psychometric cut-point that would identify patients having a suicidal emergency. SETTINGS AND DESIGN: The Emergency Department (ED) and the out-patient clinic of a university hospital. MATERIALS AND METHODS: We used the SSI to contrast 23 stable, depressed adult out-patients versus 11 depressed adult ED patients awaiting psychiatric admission for a suicidal emergency. STATISTICAL ANALYSIS: The performance of the SSI was examined with nominal logistic regression. RESULTS: ED patients were older than out-patients (P<0.001), with proportionally more men (P<0.05), and were more ethnically diverse than the outpatients (P<0.005). Compared to out-patients, ED patients were more depressed (Patient Health Questionnaire-9 score 23.1±3.8 vs. 11.7±7.3, P<0.005) and reported a greater degree of suicidal ideation (SSI scores 25.7±7.3 vs. 4.2±8.4, P<0.0001). Nominal logistic regression for the univariate model of SSI score and group yielded a score of 16 (P<0.0001) as the best cut-point in separating groups, with a corresponding Receiver Operating Characteristic Area Under the Curve = 0.94. Of 34 patients in the total sample, only two were misclassified by SSI score = 16, with both of these being false positive for ED status. Thus, the sensitivity of the cut-point was 100% with specificity of 91%. When the model was expanded to include SSI along with age, gender, ethnicity, sedative-hypnotic use, and over-the-counter use, only SSI score remained significant as a predictor. CONCLUSIONS: A SSI score ≥16 may be useful as an exclusion criterion for out-patients in depression clinical trials.

Nightmares and dysfunctional beliefs about sleep mediate the effect of insomnia symptoms on suicidal ideation.

STUDY OBJECTIVES: Many studies have reported a positive association between sleep problems and suicidal ideation. Some prospective studies in the elderly have shown that insomnia is a risk factor for suicide death after controlling for other depressive symptoms. However, hypotheses to explain how this risk is mediated have not previously been assessed. We tested the hypothesis that insomnia symptoms are related to suicidal ideation through mediation by dysfunctional beliefs and attitudes about sleep and/or nightmares. METHODS: We measured symptoms of depression, hopelessness, insomnia severity, dysfunctional beliefs and attitudes about sleep, nightmares, and suicidal ideation intensity on a convenience sample of 50 patients with depressive disorders, including 23 outpatients, 16 inpatients, and 11 suicidal ED patients. Mediation analysis was used to assess the indirect effects of insomnia symptoms on suicidal ideation through dysfunctional beliefs about sleep and through nightmares. RESULTS: Our findings again confirmed a positive association between insomnia symptoms and the intensity of suicidal ideas in depressed patients (b = 0.64, 95% CI = [0.14, 1.15]). However, we extended and clarified our earlier findings by now showing that dysfunctional beliefs and attitudes about sleep as well as nightmares may mediate the association between insomnia symptoms and suicidal ideation. The indirect effects of insomnia symptoms through dysfunctional beliefs about sleep and through nightmares were 0.38 (-0.03, 0.97) and 0.35 (0.05, 0.75), respectively. CONCLUSIONS: Nightmares as well as dysfunctional beliefs and attitudes about sleep each are positively and independently related to the intensity of suicidal ideation, and the effect of insomnia symptoms appears to be mediated through these two variables.

Developmental profiles of glutamate receptors and synaptic transmission at a single synapse in the mouse auditory brainstem.

Using whole-cell recordings from presynaptic terminals and postsynaptic principal neurons in the mouse medial nucleus of the trapezoid body (MNTB), we have characterized properties of the calyx of Held synapse during the first three postnatal weeks. We observed that evoked excitatory postsynaptic currents (EPSCs) mediated by NMDA receptors (NMDAR) increased until postnatal day 11/12 (P11/12) after which they declined to very low or undetectable levels at P16. Meanwhile, EPSCs mediated by AMPA receptors (AMPAR) showed an approximate three-fold increase in amplitude. These changes were paralleled by NMDAR and AMPAR currents evoked by exogenous NMDA and kainate to MNTB neurons except that whole-cell kainate currents remained constant after P7/8 while AMPAR-EPSCs continued to increase. We found that the decay time constant tau for NMDAR-EPSCs and AMPAR-EPSCs declined by about 30 % and 70 %, respectively. Analyses of NMDAR-EPSCs with subunit-specific pharmacological agents including ifenprodil, N,N,N',N'-tetrakis(2-pyridylmethyl)-ethylenediamine (TPEN), zinc and Mg(2+) revealed subtle developmental changes in subunit composition. As maturation progressed, this synapse displayed a reduction in the number of presynaptic spike failures and the extent of synaptic depression in response to trains of stimuli (50-300 Hz) while the recovery rate from depression accelerated. These results demonstrate profound changes in the size and kinetics of postsynaptic glutamate receptors and in the spike-firing capability of presynaptic terminals at the calyx of Held-MNTB synapse during early development. We suggest that these concurrent presynaptic and postsynaptic adaptations represent important steps for synapse consolidation and refinement and ultimately for the development of fast high-fidelity transmission at this synapse.