On Sample Based Explanation Methods for NLP:Efficiency, Faithfulness, and Semantic Evaluation

In the recent advances of natural language processing, the scale of the state-of-the-art models and datasets is usually extensive, which challenges the application of sample-based explanation methods in many aspects, such as explanation interpretability, efficiency, and faithfulness. In this work, for the first time, we can improve the interpretability of explanations by allowing arbitrary text sequences as the explanation unit. On top of this, we implement a hessian-free method with a model faithfulness guarantee. Finally, to compare our method with the others, we propose a semantic-based evaluation metric that can better align with humans' judgment of explanations than the widely adopted diagnostic or re-training measures. The empirical results on multiple real data sets demonstrate the proposed method's superior performance to popular explanation techniques such as Influence Function or TracIn on semantic evaluation.

Joint Channel Estimation and Mixed-ADCs Allocation for Massive MIMO via Deep Learning

Millimeter wave (mmWave) multi-user massive multi-input multi-output (MIMO) is a promising technique for the next generation communication systems. However, the hardware cost and power consumption grow significantly as the number of radio frequency (RF) components increases, which hampers the deployment of practical massive MIMO systems. To address this issue and further facilitate the commercialization of massive MIMO, mixed analog-to-digital converters (ADCs) architecture has been considered, where parts of conventionally assumed full-resolution ADCs are replaced by one-bit ADCs. In this paper, we first propose a deep learning-based (DL) joint pilot design and channel estimation method for mixed-ADCs mmWave massive MIMO. Specifically, we devise a pilot design neural network whose weights directly represent the optimized pilots, and develop a Runge-Kutta model-driven densely connected network as the channel estimator. Instead of randomly assigning the mixed-ADCs, we then design a novel antenna selection network for mixed-ADCs allocation to further improve the channel estimation accuracy. Moreover, we adopt an autoencoder-inspired end-to-end architecture to jointly optimize the pilot design, channel estimation and mixed-ADCs allocation networks. Simulation results show that the proposed DL-based methods have advantages over the traditional channel estimators as well as the state-of-the-art networks.

Model Aided Deep Learning Based MIMO OFDM Receiver With Nonlinear Power Amplifiers

Multi-input multi-output orthogonal frequency division multiplexing (MIMO OFDM) is a key technology for mobile communication systems. However, due to the issue of high peak-to-average power ratio (PAPR), the OFDM symbols may suffer from nonlinear distortions of the power amplifier (PA) at the transmitters, which degrades the channel estimation and detection performances of the receivers. To mitigate the clipping distortions at the receivers end, we leverage deep learning (DL) and devise a DL based receiver which is aided by the traditional least square (LS) channel estimation and the zero-forcing (ZF) equalization models. Moreover, a data driven DL based receiver without explicit channel estimation is proposed and combined with the model aided DL based receiver to further improve the performance. Simulation results showcase that the proposed model aided DL based receiver has superior performance of bit error rate and has robustness over different levels of clipping distortions.

Focus on Local: Detecting Lane Marker from Bottom Up via Key Point

Mainstream lane marker detection methods are implemented by predicting the overall structure and deriving parametric curves through post-processing. Complex lane line shapes require high-dimensional output of CNNs to model global structures, which further increases the demand for model capacity and training data. In contrast, the locality of a lane marker has finite geometric variations and spatial coverage. We propose a novel lane marker detection solution, FOLOLane, that focuses on modeling local patterns and achieving prediction of global structures in a bottom-up manner. Specifically, the CNN models lowcomplexity local patterns with two separate heads, the first one predicts the existence of key points, and the second refines the location of key points in the local range and correlates key points of the same lane line. The locality of the task is consistent with the limited FOV of the feature in CNN, which in turn leads to more stable training and better generalization. In addition, an efficiency-oriented decoding algorithm was proposed as well as a greedy one, which achieving 36% runtime gains at the cost of negligible performance degradation. Both of the two decoders integrated local information into the global geometry of lane markers. In the absence of a complex network architecture design, the proposed method greatly outperforms all existing methods on public datasets while achieving the best state-of-the-art results and real-time processing simultaneously.

Joint-DetNAS: Upgrade Your Detector with NAS, Pruning and Dynamic Distillation

We propose Joint-DetNAS, a unified NAS framework for object detection, which integrates 3 key components: Neural Architecture Search, pruning, and Knowledge Distillation. Instead of naively pipelining these techniques, our Joint-DetNAS optimizes them jointly. The algorithm consists of two core processes: student morphism optimizes the student's architecture and removes the redundant parameters, while dynamic distillation aims to find the optimal matching teacher. For student morphism, weight inheritance strategy is adopted, allowing the student to flexibly update its architecture while fully utilize the predecessor's weights, which considerably accelerates the search; To facilitate dynamic distillation, an elastic teacher pool is trained via integrated progressive shrinking strategy, from which teacher detectors can be sampled without additional cost in subsequent searches. Given a base detector as the input, our algorithm directly outputs the derived student detector with high performance without additional training. Experiments demonstrate that our Joint-DetNAS outperforms the naive pipelining approach by a great margin. Given a classic R101-FPN as the base detector, Joint-DetNAS is able to boost its mAP from 41.4 to 43.9 on MS COCO and reduce the latency by 47%, which is on par with the SOTA EfficientDet while requiring less search cost. We hope our proposed method can provide the community with a new way of jointly optimizing NAS, KD and pruning.

Project CodeNet: A Large-Scale AI for Code Dataset for Learning a Diversity of Coding Tasks

Advancements in deep learning and machine learning algorithms have enabled breakthrough progress in computer vision, speech recognition, natural language processing and beyond. In addition, over the last several decades, software has been built into the fabric of every aspect of our society. Together, these two trends have generated new interest in the fast-emerging research area of AI for Code. As software development becomes ubiquitous across all industries and code infrastructure of enterprise legacy applications ages, it is more critical than ever to increase software development productivity and modernize legacy applications. Over the last decade, datasets like ImageNet, with its large scale and diversity, have played a pivotal role in algorithmic advancements from computer vision to language and speech understanding. In this paper, we present Project CodeNet, a first-of-its-kind, very large scale, diverse, and high-quality dataset to accelerate the algorithmic advancements in AI for Code. It consists of 14M code samples and about 500M lines of code in 55 different programming languages. Project CodeNet is not only unique in its scale, but also in the diversity of coding tasks it can help benchmark: from code similarity and classification for advances in code recommendation algorithms, and code translation between a large variety programming languages, to advances in code performance (both runtime, and memory) improvement techniques. CodeNet also provides sample input and output test sets for over 7M code samples, which can be critical for determining code equivalence in different languages. As a usability feature, we provide several preprocessing tools in Project CodeNet to transform source codes into representations that can be readily used as inputs into machine learning models.

TCL: Transformer-based Dynamic Graph Modelling via Contrastive Learning

Dynamic graph modeling has recently attracted much attention due to its extensive applications in many real-world scenarios, such as recommendation systems, financial transactions, and social networks. Although many works have been proposed for dynamic graph modeling in recent years, effective and scalable models are yet to be developed. In this paper, we propose a novel graph neural network approach, called TCL, which deals with the dynamically-evolving graph in a continuous-time fashion and enables effective dynamic node representation learning that captures both the temporal and topology information. Technically, our model contains three novel aspects. First, we generalize the vanilla Transformer to temporal graph learning scenarios and design a graph-topology-aware transformer. Secondly, on top of the proposed graph transformer, we introduce a two-stream encoder that separately extracts representations from temporal neighborhoods associated with the two interaction nodes and then utilizes a co-attentional transformer to model inter-dependencies at a semantic level. Lastly, we are inspired by the recently developed contrastive learning and propose to optimize our model by maximizing mutual information (MI) between the predictive representations of two future interaction nodes. Benefiting from this, our dynamic representations can preserve high-level (or global) semantics about interactions and thus is robust to noisy interactions. To the best of our knowledge, this is the first attempt to apply contrastive learning to representation learning on dynamic graphs. We evaluate our model on four benchmark datasets for interaction prediction and experiment results demonstrate the superiority of our model.

Billion-scale Pre-trained E-commerce Product Knowledge Graph Model

In recent years, knowledge graphs have been widely applied to organize data in a uniform way and enhance many tasks that require knowledge, for example, online shopping which has greatly facilitated people's life. As a backbone for online shopping platforms, we built a billion-scale e-commerce product knowledge graph for various item knowledge services such as item recommendation. However, such knowledge services usually include tedious data selection and model design for knowledge infusion, which might bring inappropriate results. Thus, to avoid this problem, we propose a Pre-trained Knowledge Graph Model (PKGM) for our billion-scale e-commerce product knowledge graph, providing item knowledge services in a uniform way for embedding-based models without accessing triple data in the knowledge graph. Notably, PKGM could also complete knowledge graphs during servicing, thereby overcoming the common incompleteness issue in knowledge graphs. We test PKGM in three knowledge-related tasks including item classification, same item identification, and recommendation. Experimental results show PKGM successfully improves the performance of each task.

ScaleCom: Scalable Sparsified Gradient Compression for Communication-Efficient Distributed Training

Large-scale distributed training of Deep Neural Networks (DNNs) on state-of-the-art platforms is expected to be severely communication constrained. To overcome this limitation, numerous gradient compression techniques have been proposed and have demonstrated high compression ratios. However, most existing methods do not scale well to large scale distributed systems (due to gradient build-up) and/or fail to evaluate model fidelity (test accuracy) on large datasets. To mitigate these issues, we propose a new compression technique, Scalable Sparsified Gradient Compression (ScaleCom), that leverages similarity in the gradient distribution amongst learners to provide significantly improved scalability. Using theoretical analysis, we show that ScaleCom provides favorable convergence guarantees and is compatible with gradient all-reduce techniques. Furthermore, we experimentally demonstrate that ScaleCom has small overheads, directly reduces gradient traffic and provides high compression rates (65-400X) and excellent scalability (up to 64 learners and 8-12X larger batch sizes over standard training) across a wide range of applications (image, language, and speech) without significant accuracy loss.

Points as Queries: Weakly Semi-supervised Object Detection by Points

We propose a novel point annotated setting for the weakly semi-supervised object detection task, in which the dataset comprises small fully annotated images and large weakly annotated images by points. It achieves a balance between tremendous annotation burden and detection performance. Based on this setting, we analyze existing detectors and find that these detectors have difficulty in fully exploiting the power of the annotated points. To solve this, we introduce a new detector, Point DETR, which extends DETR by adding a point encoder. Extensive experiments conducted on MS-COCO dataset in various data settings show the effectiveness of our method. In particular, when using 20% fully labeled data from COCO, our detector achieves a promising performance, 33.3 AP, which outperforms a strong baseline (FCOS) by 2.0 AP, and we demonstrate the point annotations bring over 10 points in various AR metrics.