Adversarial Feature Alignment: Avoid Catastrophic Forgetting in Incremental Task Lifelong Learning.

Humans are able to master a variety of knowledge and skills with ongoing learning. By contrast, dramatic performance degradation is observed when new tasks are added to an existing neural network model. This phenomenon, termed catastrophic forgetting, is one of the major roadblocks that prevent deep neural networks from achieving human-level artificial intelligence. Several research efforts (e.g., lifelong or continual learning algorithms) have proposed to tackle this problem. However, they either suffer from an accumulating drop in performance as the task sequence grows longer, or require storing an excessive number of model parameters for historical memory, or cannot obtain competitive performance on the new tasks. In this letter, we focus on the incremental multitask image classification scenario. Inspired by the learning process of students, who usually decompose complex tasks into easier goals, we propose an adversarial feature alignment method to avoid catastrophic forgetting. In our design, both the low-level visual features and high-level semantic features serve as soft targets and guide the training process in multiple stages, which provide sufficient supervised information of the old tasks and help to reduce forgetting. Due to the knowledge distillation and regularization phenomena, the proposed method gains even better performance than fine-tuning on the new tasks, which makes it stand out from other methods. Extensive experiments in several typical lifelong learning scenarios demonstrate that our method outperforms the state-of-the-art methods in both accuracy on new tasks and performance preservation on old tasks.

NLRP11 disrupts MAVS signalosome to inhibit type I interferon signaling and virus-induced apoptosis.

MAVS signalosome plays an important role in RIG-I-like receptor (RLR)-induced antiviral signaling. Upon the recognition of viral RNAs, RLRs activate MAVS, which further recruits TRAF6 and other signaling proteins to initiate type I interferon (IFN) activation. MAVS signalosome also regulates virus-induced apoptosis to limit viral replication. However, the mechanisms that control the activity of MAVS signalosome are still poorly defined. Here, we report NLRP11, a Nod-like receptor, is induced by type I IFN and translocates to mitochondria to interact with MAVS upon viral infection. Using MAVS as a platform, NLRP11 degrades TRAF6 to attenuate the production of type I IFNs as well as virus-induced apoptosis. Our findings reveal the regulatory role of NLRP11 in antiviral immunity by disrupting MAVS signalosome.

Comparison of user satisfaction and image quality of fixed and mobile camera systems for 3-dimensional image capture of edentulous patients: A pilot clinical study.

STATEMENT OF PROBLEM: An evaluation of user satisfaction and image quality of a novel handheld purpose-built mobile camera system for 3-dimensional (3D) facial acquisition is lacking. PURPOSE: The purpose of this pilot clinical study was to assess and compare the effectiveness between a handheld mobile camera system designed for facial acquisition and a fixed static camera arrangement by comparing the time effectiveness and the operator and participant preference for the 2 techniques of image capture. MATERIAL AND METHODS: Completely edentulous participants (n=12: women=7, men=5; mean age: 74.6 years) were included in this pilot study. Images were captured with and without the prostheses in situ while maintaining "serious" and "full-smile" facial expressions. Images were captured using a mobile and a static system. The working times for the participant installation and image captures were recorded. Operator and participant perceptions of the entire experience were recorded by using visual analog scale questionnaires. Nonparametric tests were used for statistical analyses (α=.05). RESULTS: The installation time was significantly shorter for the mobile system (static=24 ±13 seconds; mobile=10 ±10 seconds), but the differences in the image capture times were not statistically significant (static: 29 ±5 seconds; mobile: 40 ±18 seconds). Operator preference was in favor of the mobile system with regard to working time (P=.002), difficulty in using (installation: P=.002; handling: P=.045), and camera weight (P=.002); however, they preferred the static arrangement for image quality (P=.003) and comfort (P=.013). The participants rated the entire photographic experience favorably, and 10 of 12 participants preferred the static camera over the mobile one. CONCLUSIONS: Despite the complexity of the installation, the static system was evaluated better for image quality; the mobile system was easier in installation and handling. The operators preferred the mobile system, and the participants preferred the static system.

LRRC14 attenuates Toll-like receptor-mediated NF-κB signaling through disruption of IKK complex.

Activation of NF-κB signaling plays pivotal roles in innate immune responses against pathogens. It requires strict control to avert inflammatory diseases. However, the mechanisms underlying this tight regulation are not completely understood. Here, we identified LRRC14, a novel member of LRR (leucine-rich repeat) protein family, as a negative regulator in TLR signaling. Expression of LRRC14 resulted in decreased activation of NF-κB, whereas knockdown of LRRC14 enhanced NF-κB activation as well as the production of inflammatory cytokines. Mechanistically, LRRC14 bound to HLH domain of IKKß to block its interaction with NEMO and thereby inhibiting the phosphorylation of IKKß and NF-κB activation. In addition, our data showed that TLR signaling led to lower expression of LRRC14. Together, LRRC14 may function as a checkpoint to prevent overzealous inflammation.

Interplay of m6A and H3K27 trimethylation restrains inflammation during bacterial infection.

While N 6-methyladenosine (m6A) is the most prevalent modification of eukaryotic messenger RNA (mRNA) involved in various cellular responses, its role in modulating bacteria-induced inflammatory response remains elusive. Here, we showed that loss of the m6A reader YTH-domain family 2 (YTHDF2) promoted demethylation of histone H3 lysine-27 trimethylation (H3K27me3), which led to enhanced production of proinflammatory cytokines and facilitated the deposition of m6A cotranscriptionally. Mechanistically, the mRNA of lysine demethylase 6B (KDM6B) was m6A-modified and its decay mediated by YTHDF2. YTHDF2 deficiency stabilized KDM6B to promote H3K27me3 demethylation of multiple proinflammatory cytokines and subsequently enhanced their transcription. Furthermore, we identified H3K27me3 as a barrier for m6A modification during transcription. KDM6B recruits the m6A methyltransferase complex to facilitate the methylation of m6A in transcribing mRNA by removing adjacent H3K27me3 barriers. These results revealed cross-talk between m6A and H3K27me3 during bacterial infection, which has broader implications for deciphering epitranscriptomics in immune homeostasis.

USP38 Couples Histone Ubiquitination and Methylation via KDM5B to Resolve Inflammation.

Chromatin modifications, such as histone acetylation, ubiquitination, and methylation, play fundamental roles in maintaining chromatin architecture and regulating gene transcription. Although their crosstalk in chromatin remodeling has been gradually uncovered, the functional relationship between histone ubiquitination and methylation in regulating immunity and inflammation remains unclear. Here, it is reported that USP38 is a novel histone deubiquitinase that works together with the histone H3K4 modifier KDM5B to orchestrate inflammatory responses. USP38 specifically removes the monoubiquitin on H2B at lysine 120, which functions as a prerequisite for the subsequent recruitment of demethylase KDM5B to the promoters of proinflammatory cytokines Il6 and Il23a during LPS stimulation. KDM5B in turn inhibits the binding of NF-κB transcription factors to the Il6 and Il23a promoters by reducing H3K4 trimethylation. Furthermore, USP38 can bind to KDM5B and prevent it from proteasomal degradation, which further enhances the function of KDM5B in the regulation of inflammation-related genes. Loss of Usp38 in mice markedly enhances susceptibility to endotoxin shock and acute colitis, and these mice display a more severe inflammatory phenotype compared to wild-type mice. The studies identify USP38-KDM5B as a distinct chromatin modification complex that restrains inflammatory responses through manipulating the crosstalk of histone ubiquitination and methylation.

Prognostic and predictive value of a microRNA signature in adults with T-cell lymphoblastic lymphoma.

New prognostic factors are needed to establish indications for haematopoietic stem cell transplantation (HSCT) in first complete remission (CR1) for T-cell lymphoblastic lymphoma (T-LBL) patients. We used microarray to compare T-LBL tissue samples (n = 75) and fetal thymus tissues (n = 20), and identified 35 differentially expressed miRNAs. Using 107 subjects as the training group, we developed a five-miRNA-based classifier to predict patient survival with LASSO Cox regression: lower risk was associated with better prognosis (disease-free survival (DFS): hazard ratio (HR) 4.548, 95% CI 2.433-8.499, p < 0.001; overall survival (OS): HR 5.030, 95% CI 2.407-10.513, p < 0.001). This classifier displayed good performance in the internal testing set (n = 106) and the independent external set (n = 304). High risk was associated with more favorable response to HSCT (DFS: HR 1.675, 95% CI 1.127-2.488, p = 0.011; OS: HR 1.602, 95% CI 1.055-2.433, p = 0.027). When combined with ECOG-PS and/or NOTCH1/FBXW7 status, this classifier had even better prognostic performance in patients receiving HSCT (DFS: HR 2.088, 95% CI 1.290-3.379, p = 0.003; OS: HR 1.996, 95% CI 1.203-3.311, p = 0.007). The five-miRNA classifier may be a useful prognostic biomarker for T-LBL adults, and could identify subjects who could benefit from HSCT.

Embryonic lethality in mice lacking Trim59 due to impaired gastrulation development.

TRIM family members have been implicated in a variety of biological processes such as differentiation and development. We here found that Trim59 plays a critical role in early embryo development from blastocyst stage to gastrula. There existed delayed development and empty yolk sacs from embryonic day (E) 8.5 in Trim59-/- embryos. No viable Trim59-/- embryos were observed beyond E9.5. Trim59 deficiency affected primary germ layer formation at the beginning of gastrulation. At E6.5 and E7.5, the expression of primary germ layer formation-associated genes including Brachyury, lefty2, Cer1, Otx2, Wnt3, and BMP4 was reduced in Trim59-/- embryos. Homozygous mutant embryonic epiblasts were contracted and the mesoderm was absent. Trim59 could interact with actin- and myosin-associated proteins. Its deficiency disturbed F-actin polymerization during inner cell mass differentiation. Trim59-mediated polymerization of F-actin was via WASH K63-linked ubiquitination. Thus, Trim59 may be a critical regulator for early embryo development from blastocyst stage to gastrula through modulating F-actin assembly.

NLRP11 attenuates Toll-like receptor signalling by targeting TRAF6 for degradation via the ubiquitin ligase RNF19A.

The adaptor protein TRAF6 has a central function in Toll-like receptor (TLR) signalling, yet the molecular mechanisms controlling its activity and stability are unclear. Here we show that NLRP11, a primate specific gene, inhibits TLR signalling by targeting TRAF6 for degradation. NLRP11 recruits the ubiquitin ligase RNF19A to catalyze K48-linked ubiquitination of TRAF6 at multiple sites, thereby leading to the degradation of TRAF6. Furthermore, deficiency in either NLRP11 or RNF19A abrogates K48-linked ubiquitination and degradation of TRAF6, which promotes activation of NF-κB and MAPK signalling and increases the production of proinflammatory cytokines. Therefore, our findings identify NLRP11 as a conserved negative regulator of TLR signalling in primate cells and reveal a mechanism by which the NLRP11-RNF19A axis targets TRAF6 for degradation.

Fusing multiview and photometric stereo for 3D reconstruction under uncalibrated illumination.

We propose a method to obtain a complete and accurate 3D model from multiview images captured under a variety of unknown illuminations. Based on recent results showing that for Lambertian objects, general illumination can be approximated well using low-order spherical harmonics, we develop a robust alternating approach to recover surface normals. Surface normals are initialized using a multi-illumination multiview stereo algorithm, then refined using a robust alternating optimization method based on the l(1) metric. Erroneous normal estimates are detected using a shape prior. Finally, the computed normals are used to improve the preliminary 3D model. The reconstruction system achieves watertight and robust 3D reconstruction while neither requiring manual interactions nor imposing any constraints on the illumination. Experimental results on both real world and synthetic data show that the technique can acquire accurate 3D models for Lambertian surfaces, and even tolerates small violations of the Lambertian assumption.