United States

Convolutional neural network (CNN) models for computer vision are powerful but lack explainability in their most basic form. This deficiency remains a key challenge when applying CNNs in important domains. Recent work on explanations through feature importance of approximate linear models has moved from input-level features (pixels or segments) to features from mid-layer feature maps in the form of concept activation vectors (CAVs).  CAVs contain concept-level information and could be learned via clustering. In this work, we rethink the ACE algorithm of Ghorbani et~al., proposing an alternative invertible concept-based explanation (ICE) framework to overcome its shortcomings. Based on the requirements of fidelity (approximate models to target models) and interpretability (being meaningful to people), we design measurements and evaluate a range of matrix factorization methods with our framework. We find that non-negative concept activation vectors (NCAVs) from non-negative matrix factorization provide superior performance in interpretability and fidelity based on computational and human subject experiments. Our framework provides both local and global concept-level explanations for pre-trained CNN models.

AAAI 2021

Invertible Concept-Based Explanations for CNN Models with Non-Negative Concept Activation Vectors

The purpose of the AAAI conference is to promote research in artificial intelligence (AI) and scientific exchange among AI researchers, practitioners, scientists, and engineers in affiliated disciplines. 

AAAI 2022

technical paper

We introduce a new dataset, MELINDA, for Multimodal biomEdicaL experImeNt methoD clAssification. The dataset is collected in a fully automated distant supervision manner, where the labels are obtained from an existing curated database, and the actual contents are extracted from papers associated with each of the records in the database. We benchmark various state-of-the-art NLP and computer vision models, including unimodal models which only take either caption texts or images as inputs, and multimodal models. Extensive experiments and analysis show that multimodal models, despite outperforming unimodal ones, still need improvements especially on a less-supervised way of grounding visual concepts with languages, and better transferability to low resource domains. We release our dataset and the benchmarks to facilitate future research in multimodal learning, especially to motivate targeted improvements for applications in scientific domains.

MELINDA: A Multimodal Dataset for Biomedical Experiment Method Classification

We investigate sublinear classical and quantum algorithms for matrix games, a fundamental problem in optimization and machine learning, with provable guarantees. Given a matrix, sublinear algorithms for the matrix game were previously known only for two special cases: (1) the maximizing vectors live in the L1-norm unit ball, and (2) the minimizing vectors live in either the L1- or the L2-norm unit ball. We give a sublinear classical algorithm that can interpolate smoothly between these two cases: for any fixed q between 1 and 2, we solve, within some additive error, matrix games where the minimizing vectors are in an Lq-norm unit ball. We also provide a corresponding sublinear quantum algorithm that solves the same task with a quadratic improvement in dimensions of the maximizing and minimizing vectors. Both our classical and quantum algorithms are optimal in the dimension parameters up to poly-logarithmic factors. Finally, we propose sublinear classical and quantum algorithms for the approximate CarathÃ©odory problem and the Lq-margin support vector machines as applications.

Sublinear Classical and Quantum Algorithms for General Matrix Games

Transformers are powerful for sequence modeling. Nearly all state-of-the-art language models and pre-trained language models are based on the Transformer architecture. However, it distinguishes sequential tokens only with the token position index. We hypothesize that better contextual representations can be generated from the Transformer with richer positional information. To verify this, we propose a segment-aware Transformer (Segatron), by replacing the original token position encoding with a combined position encoding of paragraph, sentence, and token. We first introduce the segment-aware mechanism to Transformer-XL, which is a popular Transformer-based language model with memory extension and relative position encoding. We find that our method can further improve the Transformer-XL base model and large model, achieving 17.1 perplexity on the WikiText-103 dataset. We further investigate the pre-training masked language modeling task with Segatron. Experimental results show that BERT pre-trained with Segatron (SegaBERT) can outperform BERT with vanilla Transformer on various NLP tasks, and outperforms RoBERTa on zero-shot sentence representation learning. Our code is available on GitHub.

Segatron: Segment-Aware Transformer for Language Modeling and Understanding

Crowdsourcing provides a practical way to obtain large amounts of labeled data at a low cost. However, the annotation quality of annotators varies considerably, which imposes new challenges in learning a high-quality model from the crowdsourced annotations. In this work, we provide a new perspective to decompose annotation noise into common noise and individual noise and differentiate the source of confusion based on instance difficulty and annotator expertise on a per-instance-annotator basis. We realize this new crowdsourcing model by an end-to-end learning solution with two types of noise adaptation layers: one is shared across annotators to capture their commonly shared confusions, and the other one is pertaining to each annotator to realize individual confusion. To recognize the source of noise in each annotation, we use an auxiliary network to choose from the two noise adaptation layers with respect to both instances and annotators. Extensive experiments on both synthesized and real-world benchmarks demonstrate the effectiveness of our proposed common noise adaptation solution.

Learning from Crowds by Modeling Common Confusions

Recent progress in cross-lingual relation and event extraction use graph convolutional networks (GCNs) with universal dependency parses to learn language-agnostic sentence representations such that models trained on one language can be applied to other languages. However, GCNs struggle to model words with long-range dependencies or are not directly connected in the dependency tree. To address these challenges, we propose to utilize the self-attention mechanism where we explicitly fuse structural information to learn the dependencies between words with different syntactic distances. We introduce GATE, a Graph Attention Transformer Encoder, and test its cross-lingual transferability on relation and event extraction tasks. We perform experiments on the ACE05 dataset that includes three typologically different languages: English, Chinese, and Arabic. The evaluation results show that GATE outperforms three recently proposed methods by a large margin. Our detailed analysis reveals that due to the reliance on syntactic dependencies, GATE produces robust representations that facilitate transfer across languages.

GATE: Graph Attention Transformer Encoder for Cross-Lingual Relation and Event Extraction

We study conversational dialog in which there are many possible responses to a given history. We present the MultiTalk Dataset, a corpus of over 320,000 sentences of written conversational dialog that balances a high branching factor (10) with several conversation turns (6) through selective branch continuation. We make multiple contributions to study dialog  generation in the highly branching setting. In order to evaluate a diverse set of generations, we propose a simple scoring algorithm, based on bipartite graph matching, to optimally incorporate a set of diverse references. We study multiple language generation tasks at different levels of predictive conversation depth, using textual attributes induced automatically from pretrained classifiers. Our culminating task is a challenging theory of mind problem, a controllable generation task which requires reasoning about the expected reaction of the listener.

MultiTalk: A Highly-Branching Dialog Testbed for Diverse Conversations

Empathy describes the capacity to feel, understand, and emotionally engage with what other people are experiencing. People have recently started to turn to online health communities to seek empathetic support when they undergo difficult situations such as suffering from a life-threatening disease, while others are there to provide empathetic support to those who need it. It is, therefore, important to detect the direction of empathy expressed in natural language. Previous studies only focus on the presence of empathy at a high-level and do not distinguish the direction of empathy that is expressed in textual messages. In this paper, we take one step further in the identification of perceived empathy from text by introducing IEMPATHIZE, a dataset of messages annotated with the direction of empathy exchanged in an online cancer network. We analyze user messages to identify the direction of empathy at a fine-grained level: seeking or providing empathy. Our dataset IEMPATHIZE serves as a challenging benchmark for studying empathy at a fine-grained level.

It Takes Two to Empathize: One to Seek and One to Provide

Influence diagrams (IDs) are graphical models for representing and reasoning with sequential decision-making problems under uncertainty. Limited memory influence diagrams (LIMIDs) model a decision-maker (DM) who forgets the history in the course of making a sequence of decisions. The standard inference task in IDs and LIMIDs is to compute the maximum expected utility (MEU), which is one of the most challenging tasks in graphical models. We present a model decomposition framework in both IDs and LIMIDs, which we call submodel decomposition that generates a tree of single-stage decision problems through a tree clustering scheme. We also develop a valuation algebra over the submodels that leads to a hierarchical message passing algorithm that propagates conditional expected utility functions over a submodel-tree as external messages. We show that the overall complexity is bounded by the maximum tree-width over the submodels, common in graphical model algorithms. Finally, we present a new method for computing upper bounds over a submodel-tree by first exponentiating the utility functions yielding a standard probabilistic graphical model as an upper bound and then applying standard variational upper bounds for the marginal MAP inference, yielding tighter upper bounds compared with state-of-the-art bounding schemes for the MEU task.

Submodel Decomposition Bounds for Influence Diagrams

Human understanding of narrative texts requires making commonsense inferences beyond what is stated in the text explicitly. A recent model, COMET, can generate such inferences along several dimensions such as pre- and post-conditions, motivations, and mental states of the participants. However, COMET was trained on short phrases, and is therefore discourse-agnostic. When presented with each sentence of a multi-sentence narrative, it might generate inferences that are inconsistent with the rest of the narrative. 

We present the task of discourse-aware commonsense inference. Given a sentence within a narrative, the goal is to generate commonsense inferences along predefined dimensions, while maintaining coherence with the rest of the narrative. Such large-scale paragraph-level annotation is hard to get and costly, so we use available sentence-level annotations to efficiently and automatically construct a distantly supervised corpus. 

Using this corpus, we train PARA-COMET, a discourse-aware model that incorporates paragraph-level information to generate coherent commonsense inferences from narratives. PARA-COMET captures both semantic knowledge pertaining to prior world knowledge, and episodic knowledge involving how current events relate to prior and future events in a narrative. Our results confirm that PARA-COMET outperforms the sentence-level baselines, particularly in generating inferences that are both coherent and novel.

Paragraph-Level Commonsense Transformers with Recurrent Memory

Several behavioral, social, and public health interventions, such as suicide/HIV prevention or community preparedness against natural disasters, leverage social network information to maximize outreach. Algorithmic influence maximization techniques have been proposed to aid with the choice of ``peer leaders'' or ``influencers'' in such interventions. Yet, traditional algorithms for influence maximization have not been designed with these interventions in mind. As a result, they may disproportionately exclude minority communities from the benefits of the intervention. This has motivated research on fair influence maximization. Existing techniques come with two major drawbacks. First, they require committing to a single fairness measure. Second, these measures are typically imposed as strict constraints leading to undesirable properties such as wastage of resources.
To address these shortcomings, we provide a principled characterization of the properties that a fair influence maximization algorithm should satisfy. In particular, we propose a framework based on social welfare theory, wherein the cardinal utilities derived by each community are aggregated using the isoelastic social welfare functions. Under this framework, the trade-off between fairness and efficiency can be controlled by a single inequality aversion design parameter. We then show under what circumstances our proposed principles can be satisfied by a welfare function. The resulting optimization problem is monotone and submodular and can be solved efficiently with optimality guarantees. Our framework encompasses as special cases leximin and proportional fairness. Extensive experiments on synthetic and real world datasets including a case study on landslide risk management demonstrate the efficacy of the proposed framework.

Downloads

Next from AAAI 2021

MELINDA: A Multimodal Dataset for Biomedical Experiment Method Classification

Similar lecture

Towards Automating Model Explanations with Certified Robustness Guarantees

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from AAAI 2021

MELINDA: A Multimodal Dataset for Biomedical Experiment Method Classification

Similar lecture

Towards Automating Model Explanations with Certified Robustness Guarantees

Downloads