Xiaoli Z. Fern

Ali Raza, E. Adrian Henle, Xiaoli Fern

Graph neural networks have become a powerful framework for learning complex structure-property relationships and fast screening of chemical compounds. Recently proposed methods have demonstrated that using 3D geometry information of the molecule along with the bonding structure can lead to more accurate prediction on a wide range of properties. A common practice is to use 3D geometries computed through density functional theory (DFT) for both training and testing of models. However, the computational time needed for DFT calculations can be prohibitively large. Moreover, many of the properties that we aim to predict can often be obtained with little or no overhead on top of the DFT calculations used to produce the 3D geometry information, voiding the need for a predictive model. To be practically useful for high-throughput chemical screening and drug discovery, it is desirable to work with 3D geometries obtained using less-accurate but much more efficient non-DFT methods. In this work we investigate the impact of using non-DFT conformations in the training and the testing of existing models and propose a data augmentation method for improving the prediction accuracy of classical forcefield-derived geometries.

Nora Belrose, Quintin Pope, Lucia Quirke et al.

The distributional simplicity bias (DSB) posits that neural networks learn low-order moments of the data distribution first, before moving on to higher-order correlations. In this work, we present compelling new evidence for the DSB by showing that networks automatically learn to perform well on maximum-entropy distributions whose low-order statistics match those of the training set early in training, then lose this ability later. We also extend the DSB to discrete domains by proving an equivalence between token $n$-gram frequencies and the moments of embedding vectors, and by finding empirical evidence for the bias in LLMs. Finally we use optimal transport methods to surgically edit the low-order statistics of one class to match those of another, and show that early-training networks treat the edited samples as if they were drawn from the target class. Code is available at https://github.com/EleutherAI/features-across-time.

Quintin Pope, Xiaoli Z. Fern

We study the problem of generating counterfactual text for a classifier as a means for understanding and debugging classification. Given a textual input and a classification model, we aim to minimally alter the text to change the model's prediction. White-box approaches have been successfully applied to similar problems in vision where one can directly optimize the continuous input. Optimization-based approaches become difficult in the language domain due to the discrete nature of text. We bypass this issue by directly optimizing in the latent space and leveraging a language model to generate candidate modifications from optimized latent representations. We additionally use Shapley values to estimate the combinatoric effect of multiple changes. We then use these estimates to guide a beam search for the final counterfactual text. We achieve favorable performance compared to recent white-box and black-box baselines using human and automatic evaluations. Ablation studies show that both latent optimization and the use of Shapley values improve success rate and the quality of the generated counterfactuals.

Xingyi Li, Wenxuan Wu, Xiaoli Z. Fern et al.

Recently, there has been a significant interest in performing convolution over irregularly sampled point clouds. Since point clouds are very different from regular raster images, it is imperative to study the generalization of the convolution networks more closely, especially their robustness under variations in scale and rotations of the input data. This paper investigates different variants of PointConv, a convolution network on point clouds, to examine their robustness to input scale and rotation changes. Of the variants we explored, two are novel and generated significant improvements. The first is replacing the multilayer perceptron based weight function with much simpler third degree polynomials, together with a Sobolev norm regularization. Secondly, for 3D datasets, we derive a novel viewpoint-invariant descriptor by utilizing 3D geometric properties as the input to PointConv, in addition to the regular 3D coordinates. We have also explored choices of activation functions, neighborhood, and subsampling methods. Experiments are conducted on the 2D MNIST & CIFAR-10 datasets as well as the 3D SemanticKITTI & ScanNet datasets. Results reveal that on 2D, using third degree polynomials greatly improves PointConv's robustness to scale changes and rotations, even surpassing traditional 2D CNNs for the MNIST dataset. On 3D datasets, the novel viewpoint-invariant descriptor significantly improves the performance as well as robustness of PointConv. We achieve the state-of-the-art semantic segmentation performance on the SemanticKITTI dataset, as well as comparable performance with the current highest framework on the ScanNet dataset among point-based approaches.

Ali Raza, Faaiq Waqar, Arni Sturluson et al.

Metal-organic framework (MOFs) are nanoporous materials that could be used to capture carbon dioxide from the exhaust gas of fossil fuel power plants to mitigate climate change. In this work, we design and train a message passing neural network (MPNN) to predict simulated CO$_2$ adsorption in MOFs. Towards providing insights into what substructures of the MOFs are important for the prediction, we introduce a soft attention mechanism into the readout function that quantifies the contributions of the node representations towards the graph representations. We investigate different mechanisms for sparse attention to ensure only the most relevant substructures are identified.

Hamed Shahbazi, Xiaoli Z. Fern, Reza Ghaeini et al.

Recent neural models for relation extraction with distant supervision alleviate the impact of irrelevant sentences in a bag by learning importance weights for the sentences. Efforts thus far have focused on improving extraction accuracy but little is known about their explainability. In this work we annotate a test set with ground-truth sentence-level explanations to evaluate the quality of explanations afforded by the relation extraction models. We demonstrate that replacing the entity mentions in the sentences with their fine-grained entity types not only enhances extraction accuracy but also improves explanation. We also propose to automatically generate "distractor" sentences to augment the bags and train the model to ignore the distractors. Evaluations on the widely used FB-NYT dataset show that our methods achieve new state-of-the-art accuracy while improving model explainability.

Hamed Shahbazi, Xiaoli Z. Fern, Reza Ghaeini et al.

We present a new local entity disambiguation system. The key to our system is a novel approach for learning entity representations. In our approach we learn an entity aware extension of Embedding for Language Model (ELMo) which we call Entity-ELMo (E-ELMo). Given a paragraph containing one or more named entity mentions, each mention is first defined as a function of the entire paragraph (including other mentions), then they predict the referent entities. Utilizing E-ELMo for local entity disambiguation, we outperform all of the state-of-the-art local and global models on the popular benchmarks by improving about 0.5\% on micro average accuracy for AIDA test-b with Yago candidate set. The evaluation setup of the training data and candidate set are the same as our baselines for fair comparison.

Reza Ghaeini, Xiaoli Z. Fern, Hamed Shahbazi et al.

Deep learning has emerged as a compelling solution to many NLP tasks with remarkable performances. However, due to their opacity, such models are hard to interpret and trust. Recent work on explaining deep models has introduced approaches to provide insights toward the model's behaviour and predictions, which are helpful for assessing the reliability of the model's predictions. However, such methods do not improve the model's reliability. In this paper, we aim to teach the model to make the right prediction for the right reason by providing explanation training and ensuring the alignment of the model's explanation with the ground truth explanation. Our experimental results on multiple tasks and datasets demonstrate the effectiveness of the proposed method, which produces more reliable predictions while delivering better results compared to traditionally trained models.

J. Walker Orr, Prasad Tadepalli, Janardhan Rao Doppa et al.

Scripts have been proposed to model the stereotypical event sequences found in narratives. They can be applied to make a variety of inferences including filling gaps in the narratives and resolving ambiguous references. This paper proposes the first formal framework for scripts based on Hidden Markov Models (HMMs). Our framework supports robust inference and learning algorithms, which are lacking in previous clustering models. We develop an algorithm for structure and parameter learning based on Expectation Maximization and evaluate it on a number of natural datasets. The results show that our algorithm is superior to several informed baselines for predicting missing events in partial observation sequences.

Reza Ghaeini, Xiaoli Z. Fern, Prasad Tadepalli

Recognizing sarcasm often requires a deep understanding of multiple sources of information, including the utterance, the conversational context, and real world facts. Most of the current sarcasm detection systems consider only the utterance in isolation. There are some limited attempts toward taking into account the conversational context. In this paper, we propose an interpretable end-to-end model that combines information from both the utterance and the conversational context to detect sarcasm, and demonstrate its effectiveness through empirical evaluations. We also study the behavior of the proposed model to provide explanations for the model's decisions. Importantly, our model is capable of determining the impact of utterance and conversational context on the model's decisions. Finally, we provide an ablation study to illustrate the impact of different components of the proposed model.

J. Walker Orr, Prasad Tadepalli, Xiaoli Fern

Detecting events and classifying them into predefined types is an important step in knowledge extraction from natural language texts. While the neural network models have generally led the state-of-the-art, the differences in performance between different architectures have not been rigorously studied. In this paper we present a novel GRU-based model that combines syntactic information along with temporal structure through an attention mechanism. We show that it is competitive with other neural network architectures through empirical evaluations under different random initializations and training-validation-test splits of ACE2005 dataset.

Reza Ghaeini, Xiaoli Z. Fern, Prasad Tadepalli

Deep learning models have achieved remarkable success in natural language inference (NLI) tasks. While these models are widely explored, they are hard to interpret and it is often unclear how and why they actually work. In this paper, we take a step toward explaining such deep learning based models through a case study on a popular neural model for NLI. In particular, we propose to interpret the intermediate layers of NLI models by visualizing the saliency of attention and LSTM gating signals. We present several examples for which our methods are able to reveal interesting insights and identify the critical information contributing to the model decisions.

Hamed Shahbazi, Xiaoli Z. Fern, Reza Ghaeini et al.

In this paper, we present a novel model for entity disambiguation that combines both local contextual information and global evidences through Limited Discrepancy Search (LDS). Given an input document, we start from a complete solution constructed by a local model and conduct a search in the space of possible corrections to improve the local solution from a global view point. Our search utilizes a heuristic function to focus more on the least confident local decisions and a pruning function to score the global solutions based on their local fitness and the global coherences among the predicted entities. Experimental results on CoNLL 2003 and TAC 2010 benchmarks verify the effectiveness of our model.

Reza Ghaeini, Xiaoli Z. Fern, Hamed Shahbazi et al.

We present a novel deep learning architecture to address the cloze-style question answering task. Existing approaches employ reading mechanisms that do not fully exploit the interdependency between the document and the query. In this paper, we propose a novel \emph{dependent gated reading} bidirectional GRU network (DGR) to efficiently model the relationship between the document and the query during encoding and decision making. Our evaluation shows that DGR obtains highly competitive performance on well-known machine comprehension benchmarks such as the Children's Book Test (CBT-NE and CBT-CN) and Who DiD What (WDW, Strict and Relaxed). Finally, we extensively analyze and validate our model by ablation and attention studies.

Reza Ghaeini, Xiaoli Z. Fern, Liang Huang et al.

Traditional event detection methods heavily rely on manually engineered rich features. Recent deep learning approaches alleviate this problem by automatic feature engineering. But such efforts, like tradition methods, have so far only focused on single-token event mentions, whereas in practice events can also be a phrase. We instead use forward-backward recurrent neural networks (FBRNNs) to detect events that can be either words or phrases. To the best our knowledge, this is one of the first efforts to handle multi-word events and also the first attempt to use RNNs for event detection. Experimental results demonstrate that FBRNN is competitive with the state-of-the-art methods on the ACE 2005 and the Rich ERE 2015 event detection tasks.

Reza Ghaeini, Sadid A. Hasan, Vivek Datla et al.

We present a novel deep learning architecture to address the natural language inference (NLI) task. Existing approaches mostly rely on simple reading mechanisms for independent encoding of the premise and hypothesis. Instead, we propose a novel dependent reading bidirectional LSTM network (DR-BiLSTM) to efficiently model the relationship between a premise and a hypothesis during encoding and inference. We also introduce a sophisticated ensemble strategy to combine our proposed models, which noticeably improves final predictions. Finally, we demonstrate how the results can be improved further with an additional preprocessing step. Our evaluation shows that DR-BiLSTM obtains the best single model and ensemble model results achieving the new state-of-the-art scores on the Stanford NLI dataset.

Zeyu You, Raviv Raich, Xiaoli Z. Fern et al.

We present a probabilistic modeling and inference framework for discriminative analysis dictionary learning under a weak supervision setting. Dictionary learning approaches have been widely used for tasks such as low-level signal denoising and restoration as well as high-level classification tasks, which can be applied to audio and image analysis. Synthesis dictionary learning aims at jointly learning a dictionary and corresponding sparse coefficients to provide accurate data representation. This approach is useful for denoising and signal restoration, but may lead to sub-optimal classification performance. By contrast, analysis dictionary learning provides a transform that maps data to a sparse discriminative representation suitable for classification. We consider the problem of analysis dictionary learning for time-series data under a weak supervision setting in which signals are assigned with a global label instead of an instantaneous label signal. We propose a discriminative probabilistic model that incorporates both label information and sparsity constraints on the underlying latent instantaneous label signal using cardinality control. We present the expectation maximization (EM) procedure for maximum likelihood estimation (MLE) of the proposed model. To facilitate a computationally efficient E-step, we propose both a chain and a novel tree graph reformulation of the graphical model. The performance of the proposed model is demonstrated on both synthetic and real-world data.

Behrouz Behmardi, Forrest Briggs, Xiaoli Z. Fern et al.

Multi-instance data, in which each object (bag) contains a collection of instances, are widespread in machine learning, computer vision, bioinformatics, signal processing, and social sciences. We present a maximum entropy (ME) framework for learning from multi-instance data. In this approach each bag is represented as a distribution using the principle of ME. We introduce the concept of confidence-constrained ME (CME) to simultaneously learn the structure of distribution space and infer each distribution. The shared structure underlying each density is used to learn from instances inside each bag. The proposed CME is free of tuning parameters. We devise a fast optimization algorithm capable of handling large scale multi-instance data. In the experimental section, we evaluate the performance of the proposed approach in terms of exact rank recovery in the space of distributions and compare it with the regularized ME approach. Moreover, we compare the performance of CME with Multi-Instance Learning (MIL) state-of-the-art algorithms and show a comparable performance in terms of accuracy with reduced computational complexity.

Yuanli Pei, Xiaoli Z. Fern, Rómer Rosales et al.

We study the problem of clustering with relative constraints, where each constraint specifies relative similarities among instances. In particular, each constraint $(x_i, x_j, x_k)$ is acquired by posing a query: is instance $x_i$ more similar to $x_j$ than to $x_k$? We consider the scenario where answers to such queries are based on an underlying (but unknown) class concept, which we aim to discover via clustering. Different from most existing methods that only consider constraints derived from yes and no answers, we also incorporate don't know responses. We introduce a Discriminative Clustering method with Relative Constraints (DCRC) which assumes a natural probabilistic relationship between instances, their underlying cluster memberships, and the observed constraints. The objective is to maximize the model likelihood given the constraints, and in the meantime enforce cluster separation and cluster balance by also making use of the unlabeled instances. We evaluated the proposed method using constraints generated from ground-truth class labels, and from (noisy) human judgments from a user study. Experimental results demonstrate: 1) the usefulness of relative constraints, in particular when don't know answers are considered; 2) the improved performance of the proposed method over state-of-the-art methods that utilize either relative or pairwise constraints; and 3) the robustness of our method in the presence of noisy constraints, such as those provided by human judgement.

Anh T. Pham, Raviv Raich, Xiaoli Z. Fern

Labeling data for classification requires significant human effort. To reduce labeling cost, instead of labeling every instance, a group of instances (bag) is labeled by a single bag label. Computer algorithms are then used to infer the label for each instance in a bag, a process referred to as instance annotation. This task is challenging due to the ambiguity regarding the instance labels. We propose a discriminative probabilistic model for the instance annotation problem and introduce an expectation maximization framework for inference, based on the maximum likelihood approach. For many probabilistic approaches, brute-force computation of the instance label posterior probability given its bag label is exponential in the number of instances in the bag. Our key contribution is a dynamic programming method for computing the posterior that is linear in the number of instances. We evaluate our methods using both benchmark and real world data sets, in the domain of bird song, image annotation, and activity recognition. In many cases, the proposed framework outperforms, sometimes significantly, the current state-of-the-art MIML learning methods, both in instance label prediction and bag label prediction.

Sicheng Xiong, Rómer Rosales, Yuanli Pei et al.

This work focuses on active learning of distance metrics from relative comparison information. A relative comparison specifies, for a data point triplet $(x_i,x_j,x_k)$, that instance $x_i$ is more similar to $x_j$ than to $x_k$. Such constraints, when available, have been shown to be useful toward defining appropriate distance metrics. In real-world applications, acquiring constraints often require considerable human effort. This motivates us to study how to select and query the most useful relative comparisons to achieve effective metric learning with minimum user effort. Given an underlying class concept that is employed by the user to provide such constraints, we present an information-theoretic criterion that selects the triplet whose answer leads to the highest expected gain in information about the classes of a set of examples. Directly applying the proposed criterion requires examining $O(n^3)$ triplets with $n$ instances, which is prohibitive even for datasets of moderate size. We show that a randomized selection strategy can be used to reduce the selection pool from $O(n^3)$ to $O(n)$, allowing us to scale up to larger-size problems. Experiments show that the proposed method consistently outperforms two baseline policies.

Forrest Briggs, Xiaoli Z. Fern, Jed Irvine

Bird sound data collected with unattended microphones for automatic surveys, or mobile devices for citizen science, typically contain multiple simultaneously vocalizing birds of different species. However, few works have considered the multi-label structure in birdsong. We propose to use an ensemble of classifier chains combined with a histogram-of-segments representation for multi-label classification of birdsong. The proposed method is compared with binary relevance and three multi-instance multi-label learning (MIML) algorithms from prior work (which focus more on structure in the sound, and less on structure in the label sets). Experiments are conducted on two real-world birdsong datasets, and show that the proposed method usually outperforms binary relevance (using the same features and base-classifier), and is better in some cases and worse in others compared to the MIML algorithms.

Ali Jalali, Javad Azimi, Xiaoli Fern et al.

The problem of optimizing unknown costly-to-evaluate functions has been studied for a long time in the context of Bayesian Optimization. Algorithms in this field aim to find the optimizer of the function by asking only a few function evaluations at locations carefully selected based on a posterior model. In this paper, we assume the unknown function is Lipschitz continuous. Leveraging the Lipschitz property, we propose an algorithm with a distinct exploration phase followed by an exploitation phase. The exploration phase aims to select samples that shrink the search space as much as possible. The exploitation phase then focuses on the reduced search space and selects samples closest to the optimizer. Considering the Expected Improvement (EI) as a baseline, we empirically show that the proposed algorithm significantly outperforms EI.

Javad Azimi, Ali Jalali, Xiaoli Fern

Bayesian Optimization aims at optimizing an unknown non-convex/concave function that is costly to evaluate. We are interested in application scenarios where concurrent function evaluations are possible. Under such a setting, BO could choose to either sequentially evaluate the function, one input at a time and wait for the output of the function before making the next selection, or evaluate the function at a batch of multiple inputs at once. These two different settings are commonly referred to as the sequential and batch settings of Bayesian Optimization. In general, the sequential setting leads to better optimization performance as each function evaluation is selected with more information, whereas the batch setting has an advantage in terms of the total experimental time (the number of iterations). In this work, our goal is to combine the strength of both settings. Specifically, we systematically analyze Bayesian optimization using Gaussian process as the posterior estimator and provide a hybrid algorithm that, based on the current state, dynamically switches between a sequential policy and a batch policy with variable batch sizes. We provide theoretical justification for our algorithm and present experimental results on eight benchmark BO problems. The results show that our method achieves substantial speedup (up to %78) compared to a pure sequential policy, without suffering any significant performance loss.

Javad Azimi, Ruofei Zhang, Yang Zhou et al.

Display advertising has been a significant source of revenue for publishers and ad networks in online advertising ecosystem. One of the main goals in display advertising is to maximize user response rate for advertising campaigns, such as click through rates (CTR) or conversion rates. Although in the online advertising industry we believe that the visual appearance of ads (creatives) matters for propensity of user response, there is no published work so far to address this topic via a systematic data-driven approach. In this paper we quantitatively study the relationship between the visual appearance and performance of creatives using large scale data in the world's largest display ads exchange system, RightMedia. We designed a set of 43 visual features, some of which are novel and some are inspired by related work. We extracted these features from real creatives served on RightMedia. We also designed and conducted a series of experiments to evaluate the effectiveness of visual features for CTR prediction, ranking and performance classification. Based on the evaluation results, we selected a subset of features that have the most important impact on CTR. We believe that the findings presented in this paper will be very useful for the online advertising industry in designing high-performance creatives. It also provides the research community with the first ever data set, initial insights into visual appearance's effect on user response propensity, and evaluation benchmarks for further study.