arXiv subjects

Chen Liang

Publications and source records attributed to Chen Liang.

At least 163 records · Page 9Linked to original sources

arXiv2022

CAMERO: Consistency Regularized Ensemble of Perturbed Language Models with Weight Sharing

Model ensemble is a popular approach to produce a low-variance and well-generalized model. However, it induces large memory and inference costs, which are often not affordable for real-world deployment. Existing work has resorted to sharing weights among models. However, when increasing the proportion of the shared weights, the resulting models tend to be similar, and the benefits of using model ensemble diminish. To retain ensemble benefits while maintaining a low memory cost, we propose a consistency-regularized ensemble learning approach based on perturbed models, named CAMERO. Specifically, we share the weights of bottom layers across all models and apply different perturbations to the hidden representations for different models, which can effectively promote the model diversity. Meanwhile, we apply a prediction consistency regularizer across the perturbed models to control the variance due to the model diversity. Our experiments using large language models demonstrate that CAMERO significantly improves the generalization performance of the ensemble model. Specifically, CAMERO outperforms the standard ensemble of 8 BERT-base models on the GLUE benchmark by 0.7 with a significantly smaller model size (114.2M vs. 880.6M).

cs.CL↗

arXiv2022

The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrink

Machine Learning (ML) workloads have rapidly grown in importance, but raised concerns about their carbon footprint. Four best practices can reduce ML training energy by up to 100x and CO2 emissions up to 1000x. By following best practices, overall ML energy use (across research, development, and production) held steady at <15% of Google's total energy use for the past three years. If the whole ML field were to adopt best practices, total carbon emissions from training would reduce. Hence, we recommend that ML papers include emissions explicitly to foster competition on more than just model quality. Estimates of emissions in papers that omitted them have been off 100x-100,000x, so publishing emissions has the added benefit of ensuring accurate accounting. Given the importance of climate change, we must get the numbers right to make certain that we work on its biggest challenges.

cs.LG↗

arXiv2022

Visual Abductive Reasoning

Abductive reasoning seeks the likeliest possible explanation for partial observations. Although abduction is frequently employed in human daily reasoning, it is rarely explored in computer vision literature. In this paper, we propose a new task and dataset, Visual Abductive Reasoning (VAR), for examining abductive reasoning ability of machine intelligence in everyday visual situations. Given an incomplete set of visual events, AI systems are required to not only describe what is observed, but also infer the hypothesis that can best explain the visual premise. Based on our large-scale VAR dataset, we devise a strong baseline model, Reasoner (causal-and-cascaded reasoning Transformer). First, to capture the causal structure of the observations, a contextualized directional position embedding strategy is adopted in the encoder, that yields discriminative representations for the premise and hypothesis. Then, multiple decoders are cascaded to generate and progressively refine the premise and hypothesis sentences. The prediction scores of the sentences are used to guide cross-sentence information flow in the cascaded reasoning procedure. Our VAR benchmarking results show that Reasoner surpasses many famous video-language models, while still being far behind human performance. This work is expected to foster future efforts in the reasoning-beyond-observation paradigm.

cs.CV↗

arXiv2022

Improving Molecular Contrastive Learning via Faulty Negative Mitigation and Decomposed Fragment Contrast

Deep learning has been a prevalence in computational chemistry and widely implemented in molecule property predictions. Recently, self-supervised learning (SSL), especially contrastive learning (CL), gathers growing attention for the potential to learn molecular representations that generalize to the gigantic chemical space. Unlike supervised learning, SSL can directly leverage large unlabeled data, which greatly reduces the effort to acquire molecular property labels through costly and time-consuming simulations or experiments. However, most molecular SSL methods borrow the insights from the machine learning community but neglect the unique cheminformatics (e.g., molecular fingerprints) and multi-level graphical structures (e.g., functional groups) of molecules. In this work, we propose iMolCLR: improvement of Molecular Contrastive Learning of Representations with graph neural networks (GNNs) in two aspects, (1) mitigating faulty negative contrastive instances via considering cheminformatics similarities between molecule pairs; (2) fragment-level contrasting between intra- and inter-molecule substructures decomposed from molecules. Experiments have shown that the proposed strategies significantly improve the performance of GNN models on various challenging molecular property predictions. In comparison to the previous CL framework, iMolCLR demonstrates an averaged 1.3% improvement of ROC-AUC on 7 classification benchmarks and an averaged 4.8% decrease of the error on 5 regression benchmarks. On most benchmarks, the generic GNN pre-trained by iMolCLR rivals or even surpasses supervised learning models with sophisticated architecture designs and engineered features. Further investigations demonstrate that representations learned through iMolCLR intrinsically embed scaffolds and functional groups that can reason molecule similarities.

cs.LG↗

arXiv2022

No Parameters Left Behind: Sensitivity Guided Adaptive Learning Rate for Training Large Transformer Models

Recent research has shown the existence of significant redundancy in large Transformer models. One can prune the redundant parameters without significantly sacrificing the generalization performance. However, we question whether the redundant parameters could have contributed more if they were properly trained. To answer this question, we propose a novel training strategy that encourages all parameters to be trained sufficiently. Specifically, we adaptively adjust the learning rate for each parameter according to its sensitivity, a robust gradient-based measure reflecting this parameter's contribution to the model performance. A parameter with low sensitivity is redundant, and we improve its fitting by increasing its learning rate. In contrast, a parameter with high sensitivity is well-trained, and we regularize it by decreasing its learning rate to prevent further overfitting. We conduct extensive experiments on natural language understanding, neural machine translation, and image classification to demonstrate the effectiveness of the proposed schedule. Analysis shows that the proposed schedule indeed reduces the redundancy and improves generalization performance.

cs.CL↗

arXiv2022

Event-Triggered Optimal Attitude Consensus of Multiple Rigid Body Networks with Unknown Dynamics

In this paper, an event-triggered Reinforcement Learning (RL) method is proposed for the optimal attitude consensus of multiple rigid body networks with unknown dynamics. Firstly, the consensus error is constructed through the attitude dynamics. According to the Bellman optimality principle, the implicit form of the optimal controller and the corresponding Hamilton-Jacobi-Bellman (HJB) equations are obtained. Because of the augmented system, the optimal controller can be obtained directly without relying on the system dynamics. Secondly, the self-triggered mechanism is applied to reduce the computing and communication burden when updating the controller. In order to address the problem that the HJB equations are difficult to solve analytically, a RL method which only requires measurement data at the event-triggered instants is proposed. For each agent, only one neural network is designed to approximate the optimal value function. Each neural network is updated only at the event triggered instants. Meanwhile, the Uniformly Ultimately Bounded (UUB) of the closed-loop system is obtained, and the Zeno behavior is also avoided. Finally, the simulation results on a multiple rigid body network demonstrate the validity of the proposed method.

math.OC↗

arXiv2022

TPAD: Identifying Effective Trajectory Predictions Under the Guidance of Trajectory Anomaly Detection Model

Trajectory Prediction (TP) is an important research topic in computer vision and robotics fields. Recently, many stochastic TP models have been proposed to deal with this problem and have achieved better performance than the traditional models with deterministic trajectory outputs. However, these stochastic models can generate a number of future trajectories with different qualities. They are lack of self-evaluation ability, that is, to examine the rationality of their prediction results, thus failing to guide users to identify high-quality ones from their candidate results. This hinders them from playing their best in real applications. In this paper, we make up for this defect and propose TPAD, a novel TP evaluation method based on the trajectory Anomaly Detection (AD) technique. In TPAD, we firstly combine the Automated Machine Learning (AutoML) technique and the experience in the AD and TP field to automatically design an effective trajectory AD model. Then, we utilize the learned trajectory AD model to examine the rationality of the predicted trajectories, and screen out good TP results for users. Extensive experimental results demonstrate that TPAD can effectively identify near-optimal prediction results, improving stochastic TP models' practical application effect.

cs.LG↗

arXiv2021

A General Traffic Shaping Protocol in E-Commerce

To approach different business objectives, online traffic shaping algorithms aim at improving exposures of a target set of items, such as boosting the growth of new commodities. Generally, these algorithms assume that the utility of each user-item pair can be accessed via a well-trained conversion rate prediction model. However, for real E-Commerce platforms, there are unavoidable factors preventing us from learning such an accurate model. In order to break the heavy dependence on accurate inputs of the utility, we propose a general online traffic shaping protocol for online E-Commerce applications. In our framework, we approximate the function mapping the bonus scores, which generally are the only method to influence the ranking result in the traffic shaping problem, to the numbers of exposures and purchases. Concretely, we approximate the above function by a class of the piece-wise linear function constructed on the convex hull of the explored data points. Moreover, we reformulate the online traffic shaping problem as linear programming where these piece-wise linear functions are embedded into both the objective and constraints. Our algorithm can straightforwardly optimize the linear programming in the prime space, and its solution can be simply applied by a stochastic strategy to fulfill the optimized objective and the constraints in expectation. Finally, the online A/B test shows our proposed algorithm steadily outperforms the previous industrial level traffic shaping algorithm.

cs.LG↗

arXiv2021

S+PAGE: A Speaker and Position-Aware Graph Neural Network Model for Emotion Recognition in Conversation

Emotion recognition in conversation (ERC) has attracted much attention in recent years for its necessity in widespread applications. Existing ERC methods mostly model the self and inter-speaker context separately, posing a major issue for lacking enough interaction between them. In this paper, we propose a novel Speaker and Position-Aware Graph neural network model for ERC (S+PAGE), which contains three stages to combine the benefits of both Transformer and relational graph convolution network (R-GCN) for better contextual modeling. Firstly, a two-stream conversational Transformer is presented to extract the coarse self and inter-speaker contextual features for each utterance. Then, a speaker and position-aware conversation graph is constructed, and we propose an enhanced R-GCN model, called PAG, to refine the coarse features guided by a relative positional encoding. Finally, both of the features from the former two stages are input into a conditional random field layer to model the emotion transfer.

cs.CL↗

arXiv2021

AugLiChem: Data Augmentation Library of Chemical Structures for Machine Learning

Machine learning (ML) has demonstrated the promise for accurate and efficient property prediction of molecules and crystalline materials. To develop highly accurate ML models for chemical structure property prediction, datasets with sufficient samples are required. However, obtaining clean and sufficient data of chemical properties can be expensive and time-consuming, which greatly limits the performance of ML models. Inspired by the success of data augmentations in computer vision and natural language processing, we developed AugLiChem: the data augmentation library for chemical structures. Augmentation methods for both crystalline systems and molecules are introduced, which can be utilized for fingerprint-based ML models and Graph Neural Networks(GNNs). We show that using our augmentation strategies significantly improves the performance of ML models, especially when using GNNs. In addition, the augmentations that we developed can be used as a direct plug-in module during training and have demonstrated the effectiveness when implemented with different GNN models through the AugliChem library. The Python-based package for our implementation of Auglichem: Data augmentation library for chemical structures, is publicly available at: https://github.com/BaratiLab/AugLiChem.

cs.LG↗

arXiv2021

Creative Compensation (CC): Future of Jobs with Creative Works in 3D Printing

With the continuous growth of online 3D printing community and the democratization of 3D printers, growing number of users start sharing their own 3D designs on open platforms, enabling a wide audience to search, download, and 3D print models for free. Although sharing is mostly for altruistic reasons at first, open platforms had also created potential job opportunities to compensate creative labors. This paper analyzes new job opportunities emerged in online 3D printing social platforms and patterns of seeking compensations, and reveals various motivations for posting creative content online. We find that offering exclusive membership through subscriptions, selling final products or printing services through web stores, and using affiliate links are primary means of earning profits, while there exist gaps between creators' expectations and realities. We show that various socio-economic promises emerged, leading to a win-win situation for both creators to gain extra income and audiences to have access to more quality content. We also discuss future challenges that need to be addressed, such as ethical use of opensource content.

cs.HC↗

arXiv2021

Contrastive Video-Language Segmentation

We focus on the problem of segmenting a certain object referred by a natural language sentence in video content, at the core of formulating a pinpoint vision-language relation. While existing attempts mainly construct such relation in an implicit way, i.e., grid-level multi-modal feature fusion, it has been proven problematic to distinguish semantically similar objects under this paradigm. In this work, we propose to interwind the visual and linguistic modalities in an explicit way via the contrastive learning objective, which directly aligns the referred object and the language description and separates the unreferred content apart across frames. Moreover, to remedy for the degradation problem, we present two complementary hard instance mining strategies, i.e., Language-relevant Channel Filter and Relative Hard Instance Construction. They encourage the network to exclude visual-distinguishable feature and to focus on easy-confused objects during the contrastive training. Extensive experiments on two benchmarks, i.e., A2D Sentences and J-HMDB Sentences, quantitatively demonstrate the state-of-the-arts performance of our method and qualitatively show the more accurate distinguishment between semantically similar objects over baselines.

cs.CV↗

arXiv2021

Super Tickets in Pre-Trained Language Models: From Model Compression to Improving Generalization

The Lottery Ticket Hypothesis suggests that an over-parametrized network consists of ``lottery tickets'', and training a certain collection of them (i.e., a subnetwork) can match the performance of the full model. In this paper, we study such a collection of tickets, which is referred to as ``winning tickets'', in extremely over-parametrized models, e.g., pre-trained language models. We observe that at certain compression ratios, the generalization performance of the winning tickets can not only match but also exceed that of the full model. In particular, we observe a phase transition phenomenon: As the compression ratio increases, generalization performance of the winning tickets first improves then deteriorates after a certain threshold. We refer to the tickets on the threshold as ``super tickets''. We further show that the phase transition is task and model dependent -- as the model size becomes larger and the training data set becomes smaller, the transition becomes more pronounced. Our experiments on the GLUE benchmark show that the super tickets improve single task fine-tuning by $0.9$ points on BERT-base and $1.0$ points on BERT-large, in terms of task-average score. We also demonstrate that adaptively sharing the super tickets across tasks benefits multi-task learning.

cs.LG↗

arXiv2021

A Spectral Representation of Power Systems with Applications to Adaptive Grid Partitioning and Cascading Failure Localization

Transmission line failures in power systems propagate and cascade non-locally. This well-known yet counter-intuitive feature makes it even more challenging to optimally and reliably operate these complex networks. In this work we present a comprehensive framework based on spectral graph theory that fully and rigorously captures how multiple simultaneous line failures propagate, distinguishing between non-cut and cut set outages. Using this spectral representation of power systems, we identify the crucial graph sub-structure that ensures line failure localization -- the network bridge-block decomposition. Leveraging this theory, we propose an adaptive network topology reconfiguration paradigm that uses a two-stage algorithm where the first stage aims to identify optimal clusters using the notion of network modularity and the second stage refines the clusters by means of optimal line switching actions. Our proposed methodology is illustrated using extensive numerical examples on standard IEEE networks and we discussed several extensions and variants of the proposed algorithm.

math.OC↗

arXiv2021

Carbon Emissions and Large Neural Network Training

The computation demand for machine learning (ML) has grown rapidly recently, which comes with a number of costs. Estimating the energy cost helps measure its environmental impact and finding greener strategies, yet it is challenging without detailed information. We calculate the energy use and carbon footprint of several recent large models-T5, Meena, GShard, Switch Transformer, and GPT-3-and refine earlier estimates for the neural architecture search that found Evolved Transformer. We highlight the following opportunities to improve energy efficiency and CO2 equivalent emissions (CO2e): Large but sparsely activated DNNs can consume <1/10th the energy of large, dense DNNs without sacrificing accuracy despite using as many or even more parameters. Geographic location matters for ML workload scheduling since the fraction of carbon-free energy and resulting CO2e vary ~5X-10X, even within the same country and the same organization. We are now optimizing where and when large models are trained. Specific datacenter infrastructure matters, as Cloud datacenters can be ~1.4-2X more energy efficient than typical datacenters, and the ML-oriented accelerators inside them can be ~2-5X more effective than off-the-shelf systems. Remarkably, the choice of DNN, datacenter, and processor can reduce the carbon footprint up to ~100-1000X. These large factors also make retroactive estimates of energy cost difficult. To avoid miscalculations, we believe ML papers requiring large computational resources should make energy consumption and CO2e explicit when practical. We are working to be more transparent about energy use and CO2e in our future research. To help reduce the carbon footprint of ML, we believe energy usage and CO2e should be a key metric in evaluating models, and we are collaborating with MLPerf developers to include energy usage during training and inference in this industry standard benchmark.

cs.LG↗

arXiv2021

Line Failure Localization of Power Networks Part I: Non-cut Outages

Transmission line failures in power systems propagate non-locally, making the control of the resulting outages extremely difficult. In this work, we establish a mathematical theory that characterizes the patterns of line failure propagation and localization in terms of network graph structure. It provides a novel perspective on distribution factors that precisely captures Kirchhoff's Law in terms of topological structures. Our results show that the distribution of specific collections of subtrees of the transmission network plays a critical role on the patterns of power redistribution, and motivates the block decomposition of the transmission network as a structure to understand long-distance propagation of disturbances. In Part I of this paper, we present the case when the post-contingency network remains connected after an initial set of lines are disconnected simultaneously. In Part II, we present the case when an outage separates the network into multiple islands.

eess.SY↗

arXiv2021

Line Failure Localization of Power Networks Part II: Cut Set Outages

Transmission line failure in power systems prop-agate non-locally, making the control of the resulting outages extremely difficult. In Part II of this paper, we continue the study of line failure localizability in transmission networks and characterize the impact of cut set outages. We establish a Simple Path Criterion, showing that the propagation pattern due to bridge outages, a special case of cut set failures, are fully determined by the positions in the network of the buses that participate in load balancing. We then extend our results to general cut set outages. In contrast to non-cut outages discussed in Part I whose subsequent line failures are contained within the original blocks, cut set outages typically impact the whole network, affecting the power flows on all remaining lines. We corroborate our analytical results in both parts using the IEEE 118-bus test system, in which the failure propagation patterns exhibit a clear block-diagonal structure predicted by our theory, even when using full AC power flow equations.

eess.SY↗

arXiv2021

Token-wise Curriculum Learning for Neural Machine Translation

Existing curriculum learning approaches to Neural Machine Translation (NMT) require sampling sufficient amounts of "easy" samples from training data at the early training stage. This is not always achievable for low-resource languages where the amount of training data is limited. To address such limitation, we propose a novel token-wise curriculum learning approach that creates sufficient amounts of easy samples. Specifically, the model learns to predict a short sub-sequence from the beginning part of each target sentence at the early stage of training, and then the sub-sequence is gradually expanded as the training progresses. Such a new curriculum design is inspired by the cumulative effect of translation errors, which makes the latter tokens more difficult to predict than the beginning ones. Extensive experiments show that our approach can consistently outperform baselines on 5 language pairs, especially for low-resource languages. Combining our approach with sentence-level methods further improves the performance on high-resource languages.

cs.CL↗