We use concept-based interpretable models to mitigate shortcut learning. Existing methods lack interpretability. Beginning with a Blackbox, we iteratively carve out a mixture of interpretable experts (MoIE) and a residual network. Each expert explains a subset of data using First Order Logic (FOL). While explaining a sample, the FOL from biased BB-derived MoIE detects the shortcut effectively. Finetuning the BB with Metadata Normalization (MDN) eliminates the shortcut. The FOLs from the finetuned-BB-derived MoIE verify the elimination of the shortcut. Our experiments show that MoIE does not hurt the accuracy of the original BB and eliminates shortcuts effectively.

ML model design either starts with an interpretable model or a Blackbox and explains it post hoc. Blackbox models are flexible but difficult to explain, while interpretable models are inherently explainable. Yet, interpretable models require extensive ML knowledge and tend to be less flexible and underperforming than their Blackbox variants. This paper aims to blur the distinction between a post hoc explanation of a Blackbox and constructing interpretable models. Beginning with a Blackbox, we iteratively carve out a mixture of interpretable experts (MoIE) and a residual network. Each interpretable model specializes in a subset of samples and explains them using First Order Logic (FOL), providing basic reasoning on concepts from the Blackbox. We route the remaining samples through a flexible residual. We repeat the method on the residual network until all the interpretable models explain the desired proportion of data. Our extensive experiments show that our route, interpret, and repeat approach (1) identifies a diverse set of instance-specific concepts with high concept completeness via MoIE without compromising in performance, (2) identifies the relatively "harder" samples to explain via residuals, (3) outperforms the interpretable by-design models by significant margins during test-time interventions, and (4) fixes the shortcut learned by the original Blackbox. The code for MoIE is publicly available at: https://github.com/batmanlab/ICML-2023-Route-interpret-repeat.

A highly accurate but overconfident model is ill-suited for deployment in critical applications such as healthcare and autonomous driving. The classification outcome should reflect a high uncertainty on ambiguous in-distribution samples that lie close to the decision boundary. The model should also refrain from making overconfident decisions on samples that lie far outside its training distribution, far-out-of-distribution (far-OOD), or on unseen samples from novel classes that lie near its training distribution (near-OOD). This paper proposes an application of counterfactual explanations in fixing an over-confident classifier. Specifically, we propose to fine-tune a given pre-trained classifier using augmentations from a counterfactual explainer (ACE) to fix its uncertainty characteristics while retaining its predictive performance. We perform extensive experiments with detecting far-OOD, near-OOD, and ambiguous samples. Our empirical results show that the revised model have improved uncertainty measures, and its performance is competitive to the state-of-the-art methods.

One of the challenges faced by conversational agents is their inability to identify unstated presumptions of their users' commands, a task trivial for humans due to their common sense. In this paper, we propose a zeroshot commonsense reasoning system for conversational agents in an attempt to achieve this. Our reasoner uncovers unstated presumptions from user commands satisfying a general template of if-(state), then-(action), because-(goal). Our reasoner uses a state-ofthe-art transformer-based generative commonsense knowledge base (KB) as its source of background knowledge for reasoning. We propose a novel and iterative knowledge query mechanism to extract multi-hop reasoning chains from the neural KB which uses symbolic logic rules to significantly reduce the search space. Similar to any KBs gathered to date, our commonsense KB is prone to missing knowledge. Therefore, we propose to conversationally elicit the missing knowledge from human users with our novel dynamic question generation strategy, which generates and presents contextualized queries to human users. We evaluate the model with a user study with human users that achieves a 35% higher success rate compared to SOTA.

Improving the quality of Natural Language Understanding (NLU) models, and more specifically, task-oriented semantic parsing models, in production is a cumbersome task. In this work, we present a system called AutoNLU, which we designed to scale the NLU quality improvement process. It adds automation to three key steps: detection, attribution, and correction of model errors, i.e., bugs. We detected four times more failed tasks than with random sampling, finding that even a simple active learning sampling method on an uncalibrated model is surprisingly effective for this purpose. The AutoNLU tool empowered linguists to fix ten times more semantic parsing bugs than with prior manual processes, auto-correcting 65% of all identified bugs.

In order for conversational AI systems to hold more natural and broad-ranging conversations, they will require much more commonsense, including the ability to identify unstated presumptions of their conversational partners. For example, in the command "If it snows at night then wake me up early because I don't want to be late for work" the speaker relies on commonsense reasoning of the listener to infer the implicit presumption that they wish to be woken only if it snows enough to cause traffic slowdowns. We consider here the problem of understanding such imprecisely stated natural language commands given in the form of if-(state), then-(action), because-(goal) statements. More precisely, we consider the problem of identifying the unstated presumptions of the speaker that allow the requested action to achieve the desired goal from the given state (perhaps elaborated by making the implicit presumptions explicit). We release a benchmark data set for this task, collected from humans and annotated with commonsense presumptions. We present a neuro-symbolic theorem prover that extracts multi-hop reasoning chains, and apply it to this problem. Furthermore, to accommodate the reality that current AI commonsense systems lack full coverage, we also present an interactive conversational framework built on our neuro-symbolic system, that conversationally evokes commonsense knowledge from humans to complete its reasoning chains.

Computing devices have recently become capable of interacting with their end users via natural language. However, they can only operate within a limited "supported" domain of discourse and fail drastically when faced with an out-of-domain utterance, mainly due to the limitations of their semantic parser. In this paper, we propose a semantic parser that generalizes to out-of-domain examples by learning a general strategy for parsing an unseen utterance through adapting the logical forms of seen utterances, instead of learning to generate a logical form from scratch. Our parser maintains a memory consisting of a representative subset of the seen utterances paired with their logical forms. Given an unseen utterance, our parser works by looking up a similar utterance from the memory and adapting its logical form until it fits the unseen utterance. Moreover, we present a data generation strategy for constructing utterance-logical form pairs from different domains. Our results show an improvement of up to 68.8% on one-shot parsing under two different evaluation settings compared to the baselines.

We study compositional generalization, viz., the problem of zero-shot generalization to novel compositions of concepts in a domain. Standard neural networks fail to a large extent on compositional learning. We propose Tree Stack Memory Units (Tree-SMU) to enable strong compositional generalization. Tree-SMU is a recursive neural network with Stack Memory Units (s), a novel memory augmented neural network whose memory has a differentiable stack structure. Each SMU in the tree architecture learns to read from its stack and to write to it by combining the stacks and states of its children through gating. The stack helps capture long-range dependencies in the problem domain, thereby enabling compositional generalization. Additionally, the stack also preserves the ordering of each node's descendants, thereby retaining locality on the tree. We demonstrate strong empirical results on two mathematical reasoning benchmarks. We use four compositionality tests to assess the generalization performance of Tree-SMU and show that it enables accurate compositional generalization compared to strong baselines such as Transformers and Tree-LSTMs.

Generalization to harder compositional problem instances (a.k.a extrapolation) is challenging for standard neural networks. In contrast, recursive neural networks have the potential to achieve extrapolation because they are able to capture the compositionality of tree-structured data such as mathematical equations. However, recursive networks are prone to error propagation along trees of high depth and are unable to capture long range dependencies effectively. To overcome this, we propose Tree Stack Memory Units (Tree-SMUs), a novel memory augmented recursive neural network whose nodes consist of a differentiable stack. Each SMU cell learns to read from its stack and to write to it by combining the stacks and states of its children through gating. This architecture improves both the local and global representation of compositional data due to better expressive power and the ability to capture long-range dependencies by giving each node indirect access to its descendants. We demonstrate strong empirical results on two tasks and show that Tree-SMU enables accurate extrapolation to significantly harder instances.

Neural programming involves training neural networks to learn programs, mathematics, or logic from data. Previous works have failed to achieve good generalization performance, especially on problems and programs with high complexity or on large domains. This is because they mostly rely either on black-box function evaluations that do not capture the structure of the program, or on detailed execution traces that are expensive to obtain, and hence the training data has poor coverage of the domain under consideration. We present a novel framework that utilizes black-box function evaluations, in conjunction with symbolic expressions that define relationships between the given functions. We employ tree LSTMs to incorporate the structure of the symbolic expression trees. We use tree encoding for numbers present in function evaluation data, based on their decimal representation. We present an evaluation benchmark for this task to demonstrate our proposed model combines symbolic reasoning and function evaluation in a fruitful manner, obtaining high accuracies in our experiments. Our framework generalizes significantly better to expressions of higher depth and is able to fill partial equations with valid completions.

Differential equations are used to model numerous phenomena such as heat, electrodynamics, fluid dynamics and quantum mechanics. For example, Partial Differential Equations (PDEs) have been used for boundary control of robotic aircrafts (Paranjape et al., 2013), control of autonomous agents (Bandyopadhyay et al., 2017) or generating fluid animations (Yang et al., 2016). Although some differential equations have easy to find solutions, such differential equations do not usually emerge in real-world problems. Therefore, solving differential equations is often a bottleneck in real-world applications. It is important to solve them accurately, in a timely and cost-effective manner.

In this paper, we propose guaranteed spectral methods for learning a broad range of topic models, which generalize the popular Latent Dirichlet Allocation (LDA). We overcome the limitation of LDA to incorporate arbitrary topic correlations, by assuming that the hidden topic proportions are drawn from a flexible class of Normalized Infinitely Divisible (NID) distributions. NID distributions are generated through the process of normalizing a family of independent Infinitely Divisible (ID) random variables. The Dirichlet distribution is a special case obtained by normalizing a set of Gamma random variables. We prove that this flexible topic model class can be learned via spectral methods using only moments up to the third order, with (low order) polynomial sample and computational complexity. The proof is based on a key new technique derived here that allows us to diagonalize the moments of the NID distribution through an efficient procedure that requires evaluating only univariate integrals, despite the fact that we are handling high dimensional multivariate moments. In order to assess the performance of our proposed Latent NID topic model, we use two real datasets of articles collected from New York Times and Pubmed. Our experiments yield improved perplexity on both datasets compared with the baseline.

In this paper we propose guaranteed spectral methods for learning a broad range of topic models, which generalize the popular Latent Dirichlet Allocation (LDA). We overcome the limitation of LDA to incorporate arbitrary topic correlations, by assuming that the hidden topic proportions are drawn from a flexible class of Normalized Infinitely Divisible (NID) distributions. NID distributions are generated through the process of normalizing a family of independent Infinitely Divisible (ID) random variables. The Dirichlet distribution is a special case obtained by normalizing a set of Gamma random variables. We prove that this flexible topic model class can be learnt via spectral methods using only moments up to the third order, with (low order) polynomial sample and computational complexity. The proof is based on a key new technique derived here that allows us to diagonalize the moments of the NID distribution through an efficient procedure that requires evaluating only univariate integrals, despite the fact that we are handling high dimensional multivariate moments.

Scalable probabilistic modeling and prediction in high dimensional multivariate time-series, such as dynamic social networks with co-evolving nodes and edges, is a challenging problem, particularly for systems with hidden sources of dependence and/or homogeneity. Here, we address this problem through the discovery of hierarchical latent groups. We introduce a family of Conditional Latent Tree Models (CLTM), in which tree-structured latent variables incorporate the unknown groups. The latent tree itself is conditioned on observed covariates such as seasonality, historical activity, and node attributes. We propose a statistically efficient framework for learning both the hierarchical tree structure and the parameters of the CLTM. We demonstrate competitive performance on two real world datasets, one from the students' attempts at answering questions in a psychology MOOC and the other from Twitter users participating in an emergency management discussion and interacting with one another. In addition, our modeling framework provides valuable and interpretable information about the hidden group structures and their effect on the evolution of the time series.

Author(s): Arabshahi, Forough; Huang, Furong; Anandkumar, Animashree; Butts, Carter T. | Abstract: Abstract— In this work we study scalable probabilistic modeling and prediction for predicting student performance in a Massive Open Online Course (MOOC). Students’ performance sequence form a high dimensional multivariate time series whose joint prediction is a challenging task. We solve the problem through the discovery of hierarchical latent groups that influence the dynamics of the time series. We introduce a Conditional Latent Tree Model (CLTM), in which the latent variables incorporate the unknown groups. The latent tree itself is conditioned on observed covariates such as seasonality, past activity and node attributes. We propose a statistically efficient framework for learning the hierarchical tree structure, and the parameters of the CLTM. We demonstrate competitive performance compared to the baseline (chain CRF) that does not use the hierarchical latent groupings for prediction. Our modeling framework also provides valuable and interpretable information about the hidden group structures and their effect on the evolution of the time series.Visit our project page: http://newport.eecs.uci.edu/anandkumar/Lab/Lab_sub/CLTM.html Faculty Advisor: Animashree Anandkumar, Carter T. ButtsData Science Initiative, University of California Irvine, May 2015

We consider modeling dynamic social networks with co-evolving network participants and social ties. We consider a semiparametric conditional latent tree model for vertex presence that incorporates both hierarchical latent groups and the effects of covariates such as seasonality, previous node state, and so on. For edge evolution, we assume a naive Bayes model that includes both covariate effects and the inferred states of the vertices and hidden variables. The vertex model is learned efficiently using latent tree algorithms while the edge model is learnt via logistic regression. We observe a significant improvement in performance on challenging real-world datasets of Twitter users and a network of beach attendees, with highly sparse vertex and edge presence.

In this paper a new imaging technique for early stage ultra wideband (UWB) microwave breast cancer detection is proposed. A circular array of antennas illuminates the breast tissue with UWB pulses and the backscattered signals are then passed through a beamformer designed and applied in frequency domain. This design enables the beamformer to compensate for non-integer delays and frequency dependent dispersion and at the same time increases the accuracy of the beamformer. It is shown that the proposed imaging algorithm reduces the computational cost and memory of the imaging system by decreasing the sampling rate to the Nyquist rate and significantly reducing the number of required matrix inversions. Furthermore, the proposed algorithm significantly improves the quality of the obtained image and on average the signal-to-clutter ratio of the image is increased by 89.29% compared to other cases.