Federated learning (FL) has gain growing interests for its capability of learning from distributed data sources collectively without the need of accessing the raw data samples across different sources. So far FL research has mostly focused on improvi
ng the performance, how the algorithmic disparity will be impacted for the model learned from FL and the impact of algorithmic disparity on the utility inconsistency are largely unexplored. In this paper, we propose an FL framework to jointly consider performance consistency and algorithmic fairness across different local clients (data sources). We derive our framework from a constrained multi-objective optimization perspective, in which we learn a model satisfying fairness constraints on all clients with consistent performance. Specifically, we treat the algorithm prediction loss at each local client as an objective and maximize the worst-performing client with fairness constraints through optimizing a surrogate maximum function with all objectives involved. A gradient-based procedure is employed to achieve the Pareto optimality of this optimization problem. Theoretical analysis is provided to prove that our method can converge to a Pareto solution that achieves the min-max performance with fairness constraints on all clients. Comprehensive experiments on synthetic and real-world datasets demonstrate the superiority that our approach over baselines and its effectiveness in achieving both fairness and consistency across all local clients.
In this paper, we focus on effective learning over a collaborative research network involving multiple clients. Each client has its own sample population which may not be shared with other clients due to privacy concerns. The goal is to learn a model
for each client, which behaves better than the one learned from its own data, through secure collaborations with other clients in the network. Due to the discrepancies of the sample distributions across different clients, it is not necessarily that collaborating with everyone will lead to the best local models. We propose a learning to collaborate framework, where each client can choose to collaborate with certain members in the network to achieve a collaboration equilibrium, where smaller collaboration coalitions are formed within the network so that each client can obtain the model with the best utility. We propose the concept of benefit graph which describes how each client can benefit from collaborating with other clients and develop a Pareto optimization approach to obtain it. Finally the collaboration coalitions can be derived from it based on graph operations. Our framework provides a new way of setting up collaborations in a research network. Experiments on both synthetic and real world data sets are provided to demonstrate the effectiveness of our method.
The separation of the connected and disconnected sea partons, which were uncovered in the Euclidean path-integral formulation of the hadronic tensor, is accommodated with the CT18 parametrization of the global analysis of the parton distribution func
tions (PDFs). This is achieved with the help of the distinct small $x$ behaviors of these two sea parton components and the constraint from the lattice calculation of the ratio of the strange momentum fraction to that of the ${bar u}$ or ${bar d}$ in the disconnected insertion. This allows lattice calculations of separate flavors in both the connected and disconnected insertions to be directly compared with the global analysis results term by term.
Existing person re-identification (re-id) methods are stuck when deployed to a new unseen scenario despite the success in cross-camera person matching. Recent efforts have been substantially devoted to domain adaptive person re-id where extensive unl
abeled data in the new scenario are utilized in a transductive learning manner. However, for each scenario, it is required to first collect enough data and then train such a domain adaptive re-id model, thus restricting their practical application. Instead, we aim to explore multiple labeled datasets to learn generalized domain-invariant representations for person re-id, which is expected universally effective for each new-coming re-id scenario. To pursue practicability in real-world systems, we collect all the person re-id datasets (20 datasets) in this field and select the three most frequently used datasets (i.e., Market1501, DukeMTMC, and MSMT17) as unseen target domains. In addition, we develop DataHunter that collects over 300K+ weak annotated images named YouTube-Human from YouTube street-view videos, which joins 17 remaining full labeled datasets to form multiple source domains. On such a large and challenging benchmark called FastHuman (~440K+ labeled images), we further propose a simple yet effective Semi-Supervised Knowledge Distillation (SSKD) framework. SSKD effectively exploits the weakly annotated data by assigning soft pseudo labels to YouTube-Human to improve models generalization ability. Experiments on several protocols verify the effectiveness of the proposed SSKD framework on domain generalizable person re-id, which is even comparable to supervised learning on the target domains. Lastly, but most importantly, we hope the proposed benchmark FastHuman could bring the next development of domain generalizable person re-id algorithms.
The task of video-based commonsense captioning aims to generate event-wise captions and meanwhile provide multiple commonsense descriptions (e.g., attribute, effect and intention) about the underlying event in the video. Prior works explore the commo
nsense captions by using separate networks for different commonsense types, which is time-consuming and lacks mining the interaction of different commonsense. In this paper, we propose a Hybrid Reasoning Network (HybridNet) to endow the neural networks with the capability of semantic-level reasoning and word-level reasoning. Firstly, we develop multi-commonsense learning for semantic-level reasoning by jointly training different commonsense types in a unified network, which encourages the interaction between the clues of multiple commonsense descriptions, event-wise captions and videos. Then, there are two steps to achieve the word-level reasoning: (1) a memory module records the history predicted sequence from the previous generation processes; (2) a memory-routed multi-head attention (MMHA) module updates the word-level attention maps by incorporating the history information from the memory module into the transformer decoder for word-level reasoning. Moreover, the multimodal features are used to make full use of diverse knowledge for commonsense reasoning. Experiments and abundant analysis on the large-scale Video-to-Commonsense benchmark show that our HybridNet achieves state-of-the-art performance compared with other methods.
Benefited from considerable pixel-level annotations collected from a specific situation (source), the trained semantic segmentation model performs quite well, but fails in a new situation (target) due to the large domain shift. To mitigate the domain
gap, previous cross-domain semantic segmentation methods always assume the co-existence of source data and target data during distribution alignment. However, the access to source data in the real scenario may raise privacy concerns and violate intellectual property. To tackle this problem, we focus on an interesting and challenging cross-domain semantic segmentation task where only the trained source model is provided to the target domain, and further propose a unified framework called Domain Adaptive Semantic Segmentation without Source data (DAS$^3$ for short). Specifically, DAS$^3$ consists of three schemes, i.e., feature alignment, self-training, and information propagation. First, we mainly develop a focal entropic loss on the network outputs to implicitly align the target features with unseen source features via the provided source model. Second, besides positive pseudo labels in vanilla self-training, we first introduce negative pseudo labels to the field and develop a bi-directional self-training strategy to enhance the representation learning in the target domain. Finally, the information propagation scheme further reduces the intra-domain discrepancy within the target domain via pseudo semi-supervised learning. Extensive results on synthesis-to-real and cross-city driving datasets validate DAS$^3$ yields state-of-the-art performance, even on par with methods that need access to source data.
To alleviate the burden of labeling, unsupervised domain adaptation (UDA) aims to transfer knowledge in previous related labeled datasets (source) to a new unlabeled dataset (target). Despite impressive progress, prior methods always need to access t
he raw source data and develop data-dependent alignment approaches to recognize the target samples in a transductive learning manner, which may raise privacy concerns from source individuals. Several recent studies resort to an alternative solution by exploiting the well-trained white-box model instead of the raw data from the source domain, however, it may leak the raw data through generative adversarial training. This paper studies a practical and interesting setting for UDA, where only a black-box source model (i.e., only network predictions are available) is provided during adaptation in the target domain. Besides, different neural networks are even allowed to be employed for different domains. For this new problem, we propose a novel two-step adaptation framework called Distill and Fine-tune (Dis-tune). Specifically, Dis-tune first structurally distills the knowledge from the source model to a customized target model, then unsupervisedly fine-tunes the distilled model to fit the target domain. To verify the effectiveness, we consider two UDA scenarios (ie, closed-set and partial-set), and discover that Dis-tune achieves highly competitive performance to state-of-the-art approaches.
We compute the overlap Dirac spectrum on three ensembles generated using 2+1 flavor domain wall fermions. The spectral density is determined up to $lambdasim$100 MeV with sub-percentage statistical uncertainty. The three ensembles have different latt
ice spacings and two of them have quark masses tuned to the physical point. We show that we can resolve the flavor content of the sea quarks and constrain their masses using the Dirac spectral density. We find that the density is close to a constant below $lambdale$ 20 MeV (but 10% higher than that in the 2-flavor chiral limit) as predicted by chiral perturbative theory ($chi$PT), and then increases linearly due to the strange quark mass. Using the next to leading order $chi$PT, one can extract the light and strange quark masses with $sim$20% uncertainties. Using the non-perturbative RI/MOM renormalization, we obtain the chiral condensates at $overline{textrm{MS}}$ 2 GeV as $Sigma=(260.3(0.7)(1.3)(0.7)(0.8) textrm{MeV})^3$ in the $N_f=2$ (keeping the strange quark mass at the physical point) chiral limit and $Sigma_0=(232.6(0.9)(1.2)(0.7)(0.8) textrm{MeV})^3$ in the $N_f=3$ chiral limit, where the four uncertainties come from the statistical fluctuation, renormalization constant, continuum extrapolation and lattice spacing determination. Note that {$Sigma/Sigma_0=1.40(2)(2)$ is much larger than 1} due to the strange quark mass effect.
Recent studies show that crowd-sourced Natural Language Inference (NLI) datasets may suffer from significant biases like annotation artifacts. Models utilizing these superficial clues gain mirage advantages on the in-domain testing set, which makes t
he evaluation results over-estimated. The lack of trustworthy evaluation settings and benchmarks stalls the progress of NLI research. In this paper, we propose to assess a models trustworthy generalization performance with cross-datasets evaluation. We present a new unified cross-datasets benchmark with 14 NLI datasets, and re-evaluate 9 widely-used neural network-based NLI models as well as 5 recently proposed debiasing methods for annotation artifacts. Our proposed evaluation scheme and experimental baselines could provide a basis to inspire future reliable NLI research.
Authentication is the task of confirming the matching relationship between a data instance and a given identity. Typical examples of authentication problems include face recognition and person re-identification. Data-driven authentication could be af
fected by undesired biases, i.e., the models are often trained in one domain (e.g., for people wearing spring outfits) while applied in other domains (e.g., they change the clothes to summer outfits). Previous works have made efforts to eliminate domain-difference. They typically assume domain annotations are provided, and all the domains share classes. However, for authentication, there could be a large number of domains shared by different identities/classes, and it is impossible to annotate these domains exhaustively. It could make domain-difference challenging to model and eliminate. In this paper, we propose a domain-agnostic method that eliminates domain-difference without domain labels. We alternately perform latent domain discovery and domain-difference elimination until our model no longer detects domain-difference. In our approach, the latent domains are discovered by learning the heterogeneous predictive relationships between inputs and outputs. Then domain-difference is eliminated in both class-dependent and class-independent spaces to improve robustness of elimination. We further extend our method to a meta-learning framework to pursue more thorough domain-difference elimination. Comprehensive empirical evaluation results are provided to demonstrate the effectiveness and superiority of our proposed method.
