A Domain Adaptive Person Re-Identification Based on Dual Attention Mechanism and Camstyle Transfer

Due to the variation in the image capturing process, the difference between source and target sets causes a challenge in unsupervised domain adaptation (UDA) on person re-identification (re-ID). Given a labeled source training set and an unlabeled target training set, this paper focuses on improving the generalization ability of the re-ID model on the target testing set. The proposed method enforces two properties at the same time: (1) camera invariance is achieved through the positive learning formed by unlabeled target images and their camera style transfer counterparts; and (2) the robustness of the backbone network feature extraction is improved, and the accuracy of feature extraction is enhanced by adding a position-channel dual attention mechanism. The proposed network model uses a classic dual-stream network. Comparative experimental results on three public benchmarks prove the superiority of the proposed method.

BEAR reveals that increased fidelity variants can successfully reduce the mismatch tolerance of adenine but not cytosine base editors

Adenine and cytosine base editors (ABE, CBE) allow for precision genome engineering. Here, Base Editor Activity Reporter (BEAR), a plasmid-based fluorescent tool is introduced, which can be applied to report on ABE and CBE editing in a virtually unrestricted sequence context or to label base edited cells for enrichment. Using BEAR-enrichment, we increase the yield of base editing performed by nuclease inactive base editors to the level of the nickase versions while maintaining significantly lower indel background. Furthermore, by exploiting the semi-high-throughput potential of BEAR, we examine whether increased fidelity SpCas9 variants can be used to decrease SpCas9-dependent off-target effects of ABE and CBE. Comparing them on the same target sets reveals that CBE remains active on sequences, where increased fidelity mutations and/or mismatches decrease the activity of ABE. Our results suggest that the deaminase domain of ABE is less effective to act on rather transiently separated target DNA strands, than that of CBE explaining its lower mismatch tolerance.

The YTHDF proteins ECT2 and ECT3 bind largely overlapping target sets and influence target mRNA abundance, not alternative polyadenylation

Gene regulation via N6-methyladenosine (m6A) in mRNA involves RNA-binding proteins that recognize m6A via a YT521-B homology (YTH) domain. The plant YTH domain proteins ECT2 and ECT3 act genetically redundantly in stimulating cell proliferation during organogenesis, but several fundamental questions regarding their mode of action remain unclear. Here, we use HyperTRIBE (targets of RNA-binding proteins identified by editing) to show that most ECT2 and ECT3 targets overlap, with only few examples of preferential targeting by either of the two proteins. HyperTRIBE in different mutant backgrounds also provides direct views of redundant and specific target interactions of the two proteins. We also show that contrary to conclusions of previous reports, ECT2 does not accumulate in the nucleus. Accordingly, inactivation of ECT2, ECT3 and their surrogate ECT4 does not change patterns of polyadenylation site choice in ECT2/3 target mRNAs, but does lead to lower steady state accumulation of target mRNAs. In addition, mRNA and microRNA expression profiles show indications of stress response activation in ect2/ect3/ect4 mutants, likely via indirect effects. Thus, previous suggestions of control of alternative polyadenylation by ECT2 are not supported by evidence, and ECT2 and ECT3 act largely redundantly to regulate target mRNA, including its abundance, in the cytoplasm.

The Arabidopsis m6A-binding proteins ECT2 and ECT3 bind largely overlapping mRNA target sets and influence target mRNA abundance, not alternative polyadenylation

Gene regulation via N6-methyladenosine (m6A) in mRNA involves RNA-binding proteins that recognize m6A via a YT521-B homology (YTH) domain. The plant YTH domain proteins ECT2 and ECT3 act genetically redundantly in stimulating cell proliferation during organogenesis, but several fundamental questions regarding their mode of action remain unclear. Here, we use HyperTRIBE (targets of RNA-binding proteins identified by editing) to show that most ECT2 and ECT3 targets overlap, with only few examples of preferential targeting by either of the two proteins. HyperTRIBE in different mutant backgrounds also provides direct views of redundant and specific target interactions of the two proteins. We also show that contrary to conclusions of previous reports, ECT2 does not accumulate in the nucleus. Accordingly, inactivation of ECT2, ECT3 and their surrogate ECT4 does not change patterns of polyadenylation site choice in ECT2/3 target mRNAs, but does lead to lower steady state accumulation of target mRNAs. In addition, mRNA and microRNA expression profiles show indications of stress response activation in ect2/ect3/ect4 mutants, likely via indirect effects. Thus, previous suggestions of control of alternative polyadenylation by ECT2 are not supported by evidence, and ECT2 and ECT3 act largely redundantly to regulate target mRNA, including its abundance, in the cytoplasm.

Deep and Wide Transfer Learning with Kernel Matching for Pooling Data from Electroencephalography and Psychological Questionnaires

Motor imagery (MI) promotes motor learning and encourages brain–computer interface systems that entail electroencephalogram (EEG) decoding. However, a long period of training is required to master brain rhythms’ self-regulation, resulting in users with MI inefficiency. We introduce a parameter-based approach of cross-subject transfer-learning to improve the performances of poor-performing individuals in MI-based BCI systems, pooling data from labeled EEG measurements and psychological questionnaires via kernel-embedding. To this end, a Deep and Wide neural network for MI classification is implemented to pre-train the network from the source domain. Then, the parameter layers are transferred to initialize the target network within a fine-tuning procedure to recompute the Multilayer Perceptron-based accuracy. To perform data-fusion combining categorical features with the real-valued features, we implement stepwise kernel-matching via Gaussian-embedding. Finally, the paired source–target sets are selected for evaluation purposes according to the inefficiency-based clustering by subjects to consider their influence on BCI motor skills, exploring two choosing strategies of the best-performing subjects (source space): single-subject and multiple-subjects. Validation results achieved for discriminant MI tasks demonstrate that the introduced Deep and Wide neural network presents competitive performance of accuracy even after the inclusion of questionnaire data.

A Camera Identity-guided Distribution Consistency Method for Unsupervised Multi-target Domain Person Re-identification

Unsupervised domain adaptation (UDA) for person re-identification (re-ID) is a challenging task due to large variations in human classes, illuminations, camera views, and so on. Currently, existing UDA methods focus on two-domain adaptation and are generally trained on one labeled source set and adapted on the other unlabeled target set. In this article, we put forward a new issue on person re-ID, namely, unsupervised multi-target domain adaptation (UMDA). It involves one labeled source set and multiple unlabeled target sets, which is more reasonable for practical real-world applications. Enabling UMDA has to learn the consistency for multiple domains, which is significantly different from the UDA problem. To ensure distribution consistency and learn the discriminative embedding, we further propose the Camera Identity-guided Distribution Consistency method that performs an alignment operation for multiple domains. The camera identities are encoded into the image semantic information to facilitate the adaptation of features. According to our knowledge, this is the first attempt on the unsupervised multi-target domain adaptation learning. Extensive experiments are executed on Market-1501, DukeMTMC-reID, MSMT17, PersonX, and CUHK03, and our method has achieved very competitive re-ID accuracy in multi-target domains against numerous state-of-the-art methods.

Multiple Capture in a Group Pursuit Problem with Fractional Derivatives and Phase Restrictions

The problem of conflict interaction between a group of pursuers and an evader in a finite-dimensional Euclidean space is considered. All participants have equal opportunities. The dynamics of all players are described by a system of differential equations with fractional derivatives in the form D(α)zi=azi+ui−v,ui,v∈V, where D(α)f is a Caputo derivative of order α of the function f. Additionally, it is assumed that in the process of the game the evader does not move out of a convex polyhedral cone. The set of admissible controls V is a strictly convex compact and a is a real number. The goal of the group of pursuers is to capture of the evader by no less than m different pursuers (the instants of capture may or may not coincide). The target sets are the origin. For such a conflict-controlled process, we derive conditions on its parameters and initial state, which are sufficient for the trajectories of the players to meet at a certain instant of time for any counteractions of the evader. The method of resolving functions is used to solve the problem, which is used in differential games of pursuit by a group of pursuers of one evader.

Missingness Adapted Group Informed Clustered (MAGIC)-LASSO: A novel paradigm for prediction in data with widespread non-random missingness

The availability of large-scale biobanks linking rich phenotypes and biological measures are a powerful opportunity for scientific discovery. However, real-world collections frequently have extensive non-random missing data. Machine learning methods are able to predict missing data but performance is significantly impaired by block-wise missingness inherent to many biobanks. To address this, we developed Missingness Adapted Group-wise Informed Clustered LASSO (MAGIC-LASSO) which performs hierarchical clustering of variables based on missingness followed by sequential Group LASSO within clusters. Variables are pre-filtered for missingness and balance between training and target sets with final models built using stepwise inclusion of features ranked by completeness. This research has been conducted using the UK Biobank (n>500k) to predict unmeasured Alcohol Use Disorders Identification Test (AUDIT.) The phenotypic correlation between measured and predicted total score was 0.67 while genetic correlations between independent subjects was >0.86, demonstrating the method has significant accuracy and utility.

Equating and Assigning Targets in the Adapted Alternating Treatments Design: Review of Special Education Journals

The current review examined the prevalence of the adapted alternating treatments design (AATD) across 22 special education journals and methods to equate and assign target sets to experimental conditions in the AATD. Since the seminal description of the design in 1985, a total of 49 articles were published using the AATD across 12 of the reviewed journals. The most prominent methods of equating target sets differed from prior reviews of behavior-analytic journals, likely due to the preponderance of response chains being targeted in special education research using the AATD. The majority of articles describe at least one method for equating target sets, although multiple methods were common. Additional methodological strengths in this literature included methods to reduce potential bias when assigning target sets to experimental conditions and counterbalancing target sets across participants. Considerations for practitioners and researchers when using the AATD are described.