Adam is the important optimization algorithm to guarantee efficiency and accuracy for training many important tasks such as BERT and ImageNet. However, Adam is generally not compatible with information (gradient) compression technology. Therefore, the communication usually becomes the bottleneck for parallelizing Adam. In this paper, we propose a communication efficient {\bf A}DAM {\bf p}reconditioned {\bf M}omentum SGD algorithm-- named APMSqueeze-- through an error compensated method compressing gradients. The proposed algorithm achieves a similar convergence efficiency to Adam in term of epochs, but significantly reduces the running time per epoch. In terms of end-to-end performance (including the full-precision pre-condition step), APMSqueeze is able to provide {sometimes by up to $2-10\times$ speed-up depending on network bandwidth.} We also conduct theoretical analysis on the convergence and efficiency.
Common spatial pattern (CSP) is a popular feature extraction method for electroencephalogram (EEG) motor imagery (MI). This study modifies the conventional CSP algorithm to improve the multi-class MI classification accuracy and ensure the computation process is efficient. The EEG MI data is gathered from the Brain-Computer Interface (BCI) Competition IV. At first, a bandpass filter and a time-frequency analysis are performed for each experiment trial. Then, the optimal EEG signals for every experiment trials are selected based on the signal energy for CSP feature extraction. In the end, the extracted features are classified by three classifiers, linear discriminant analysis (LDA), na\"ive Bayes (NVB), and support vector machine (SVM), in parallel for classification accuracy comparison. The experiment results show the proposed algorithm average computation time is 37.22% less than the FBCSP (1st winner in the BCI Competition IV) and 4.98% longer than the conventional CSP method. For the classification rate, the proposed algorithm kappa value achieved 2nd highest compared with the top 3 winners in BCI Competition IV.
Drivers cognitive and physiological states affect their ability to control their vehicles. Thus, these driver states are important to the safety of automobiles. The design of advanced driver assistance systems (ADAS) or autonomous vehicles will depend on their ability to interact effectively with the driver. A deeper understanding of the driver state is, therefore, paramount. EEG is proven to be one of the most effective methods for driver state monitoring and human error detection. This paper discusses EEG-based driver state detection systems and their corresponding analysis algorithms over the last three decades. First, the commonly used EEG system setup for driver state studies is introduced. Then, the EEG signal preprocessing, feature extraction, and classification algorithms for driver state detection are reviewed. Finally, EEG-based driver state monitoring research is reviewed in-depth, and its future development is discussed. It is concluded that the current EEG-based driver state monitoring algorithms are promising for safety applications. However, many improvements are still required in EEG artifact reduction, real-time processing, and between-subject classification accuracy.
To automate the generation of interactive features, recent methods are proposed to either explicitly traverse the interactive feature space or implicitly express the interactions via intermediate activations of some designed models. These two kinds of methods show that there is essentially a trade-off between feature interpretability and efficient search. To possess both of their merits, we propose a novel method named Feature Interaction Via Edge Search (FIVES), which formulates the task of interactive feature generation as searching for edges on the defined feature graph. We first present our theoretical evidence that motivates us to search for interactive features in an inductive manner. Then we instantiate this search strategy by alternatively updating the edge structure and the predictive model of a graph neural network (GNN) associated with the defined feature graph. In this way, the proposed FIVES method traverses a trimmed search space and enables explicit feature generation according to the learned adjacency tensor of the GNN. Experimental results on both benchmark and real-world datasets demonstrate the advantages of FIVES over several state-of-the-art methods.
Knowledge Graphs (KG) are gaining increasing attention in both academia and industry. Despite their diverse benefits, recent research have identified social and cultural biases embedded in the representations learned from KGs. Such biases can have detrimental consequences on different population and minority groups as applications of KG begin to intersect and interact with social spheres. This paper aims at identifying and mitigating such biases in Knowledge Graph (KG) embeddings. As a first step, we explore popularity bias -- the relationship between node popularity and link prediction accuracy. In case of node2vec graph embeddings, we find that prediction accuracy of the embedding is negatively correlated with the degree of the node. However, in case of knowledge-graph embeddings (KGE), we observe an opposite trend. As a second step, we explore gender bias in KGE, and a careful examination of popular KGE algorithms suggest that sensitive attribute like the gender of a person can be predicted from the embedding. This implies that such biases in popular KGs is captured by the structural properties of the embedding. As a preliminary solution to debiasing KGs, we introduce a novel framework to filter out the sensitive attribute information from the KG embeddings, which we call FAN (Filtering Adversarial Network). We also suggest the applicability of FAN for debiasing other network embeddings which could be explored in future work.
Machine learning (ML) applications have been thriving recently, largely attributed to the increasing availability of data. However, inconsistency and incomplete information are ubiquitous in real-world datasets, and their impact on ML applications remains elusive. In this paper, we present a formal study of this impact by extending the notion of Certain Answers for Codd tables, which has been explored by the database research community for decades, into the field of machine learning. Specifically, we focus on classification problems and propose the notion of "Certain Predictions" (CP) -- a test data example can be certainly predicted (CP'ed) if all possible classifiers trained on top of all possible worlds induced by the incompleteness of data would yield the same prediction. We study two fundamental CP queries: (Q1) checking query that determines whether a data example can be CP'ed; and (Q2) counting query that computes the number of classifiers that support a particular prediction (i.e., label). Given that general solutions to CP queries are, not surprisingly, hard without assumption over the type of classifier, we further present a case study in the context of nearest neighbor (NN) classifiers, where efficient solutions to CP queries can be developed -- we show that it is possible to answer both queries in linear or polynomial time over exponentially many possible worlds. We demonstrate one example use case of CP in the important application of "data cleaning for machine learning (DC for ML)." We show that our proposed CPClean approach built based on CP can often significantly outperform existing techniques in terms of classification accuracy with mild manual cleaning effort.
In this work, we present a text generation approach with multi-attribute control for data augmentation. We introduce CGA, a Variational Autoencoder architecture, to control, generate, and augment text. CGA is able to generate natural sentences with multiple controlled attributes by combining adversarial learning with a context-aware loss. The scalability of our approach is established through a single discriminator, independently of the number of attributes. As the main application of our work, we test the potential of this new model in a data augmentation use case. In a downstream NLP task, the sentences generated by our CGA model not only show significant improvements over a strong baseline, but also a classification performance very similar to real data. Furthermore, we are able to show high quality, diversity and attribute control in the generated sentences through a series of automatic and human assessments.
One of the most popular paradigms of applying large, pre-trained NLP models such as BERT is to fine-tune it on a smaller dataset. However, one challenge remains as the fine-tuned model often overfits on smaller datasets. A symptom of this phenomenon is that irrelevant words in the sentences, even when they are obvious to humans, can substantially degrade the performance of these fine-tuned BERT models. In this paper, we propose a novel technique, called Self-Supervised Attention (SSA) to help facilitate this generalization challenge. Specifically, SSA automatically generates weak, token-level attention labels iteratively by "probing" the fine-tuned model from the previous iteration. We investigate two different ways of integrating SSA into BERT and propose a hybrid approach to combine their benefits. Empirically, on a variety of public datasets, we illustrate significant performance improvement using our SSA-enhanced BERT model.
International stakeholders increasingly invest in offsetting carbon emissions, for example, via issuing Payments for Ecosystem Services (PES) to forest conservation projects. Issuing trusted payments requires a transparent monitoring, reporting, and verification (MRV) process of the ecosystem services (e.g., carbon stored in forests). The current MRV process, however, is either too expensive (on-ground inspection of forest) or inaccurate (satellite). Recent works propose low-cost and accurate MRV via automatically determining forest carbon from drone imagery, collected by the landowners. The automation of MRV, however, opens up the possibility that landowners report untruthful drone imagery. To be robust against untruthful reporting, we propose TrueBranch, a metric learning-based algorithm that verifies the truthfulness of drone imagery from forest conservation projects. TrueBranch aims to detect untruthfully reported drone imagery by matching it with public satellite imagery. Preliminary results suggest that nominal distance metrics are not sufficient to reliably detect untruthfully reported imagery. TrueBranch leverages metric learning to create a feature embedding in which truthfully and untruthfully collected imagery is easily distinguishable by distance thresholding.