Behavioral changes are the earliest signs of a mental disorder, but arguably, the dynamics of brain function gets affected even earlier. Subsequently, spatio-temporal structure of disorder-specific dynamics is crucial for early diagnosis and understanding the disorder mechanism. A common way of learning discriminatory features relies on training a classifier and evaluating feature importance. Classical classifiers, based on handcrafted features are quite powerful, but suffer the curse of dimensionality when applied to large input dimensions of spatio-temporal data. Deep learning algorithms could handle the problem and a model introspection could highlight discriminatory spatio-temporal regions but need way more samples to train. In this paper we present a novel self supervised training schema which reinforces whole sequence mutual information local to context (whole MILC). We pre-train the whole MILC model on unlabeled and unrelated healthy control data. We test our model on three different disorders (i) Schizophrenia (ii) Autism and (iii) Alzheimers and four different studies. Our algorithm outperforms existing self-supervised pre-training methods and provides competitive classification results to classical machine learning algorithms. Importantly, whole MILC enables attribution of subject diagnosis to specific spatio-temporal regions in the fMRI signal.
Deep Neural Networks in NLP have enabled systems to learn complex non-linear relationships. One of the major bottlenecks towards being able to use DNNs for real world applications is their characterization as black boxes. To solve this problem, we introduce a model agnostic algorithm which calculates phrase-wise importance of input features. We contend that our method is generalizable to a diverse set of tasks, by carrying out experiments for both Regression and Classification. We also observe that our approach is robust to outliers, implying that it only captures the essential aspects of the input.
Predictive clustering trees (PCTs) are a well established generalization of standard decision trees, which can be used to solve a variety of predictive modeling tasks, including structured output prediction. Combining them into ensembles yields state-of-the-art performance. Furthermore, the ensembles of PCTs can be interpreted by calculating feature importance scores from the learned models. However, their learning time scales poorly with the dimensionality of the output space. This is often problematic, especially in (hierarchical) multi-label classification, where the output can consist of hundreds of potential labels. Also, learning of PCTs can not exploit the sparsity of data to improve the computational efficiency, which is common in both input (molecular fingerprints, bag of words representations) and output spaces (in multi-label classification, examples are often labeled with only a fraction of possible labels). In this paper, we propose oblique predictive clustering trees, capable of addressing these limitations. We design and implement two methods for learning oblique splits that contain linear combinations of features in the tests, hence a split corresponds to an arbitrary hyperplane in the input space. The methods are efficient for high dimensional data and capable of exploiting sparse data. We experimentally evaluate the proposed methods on 60 benchmark datasets for 6 predictive modeling tasks. The results of the experiments show that oblique predictive clustering trees achieve performance on-par with state-of-the-art methods and are orders of magnitude faster than standard PCTs. We also show that meaningful feature importance scores can be extracted from the models learned with the proposed methods.
Product search is the most common way for people to satisfy their shopping needs on e-commerce websites. Products are typically annotated with one of several broad categorical tags, such as "Clothing" or "Electronics", as well as finer-grained categories like "Refrigerator" or "TV", both under "Electronics". These tags are used to construct a hierarchy of query categories. Distributions of features such as price and brand popularity vary wildly across query categories. In addition, feature importance for the purpose of CTR/CVR predictions differs from one category to another. In this work, we leverage the Mixture of Expert (MoE) framework to learn a ranking model that specializes for each query category. In particular, our gate network relies solely on the category ids extracted from the user query. While classical MoE's pick expert towers spontaneously for each input example, we explore two techniques to establish more explicit and transparent connections between the experts and query categories. To help differentiate experts on their domain specialties, we introduce a form of adversarial regularization among the expert outputs, forcing them to disagree with one another. As a result, they tend to approach each prediction problem from different angles, rather than copying one another. This is validated by a much stronger clustering effect of the gate output vectors under different categories. In addition, soft gating constraints based on the categorical hierarchy are imposed to help similar products choose similar gate values. and make them more likely to share similar experts. This allows aggregation of training data among smaller sibling categories to overcome data scarcity.
Anomaly Detection is an unsupervised learning task aimed at detecting anomalous behaviours with respect to historical data. In particular, multivariate Anomaly Detection has an important role in many applications thanks to the capability of summarizing the status of a complex system or observed phenomenon with a single indicator (typically called `Anomaly Score') and thanks to the unsupervised nature of the task that does not require human tagging. The Isolation Forest is one of the most commonly adopted algorithms in the field of Anomaly Detection, due to its proven effectiveness and low computational complexity. A major problem affecting Isolation Forest is represented by the lack of interpretability, an effect of the inherent randomness governing the splits performed by the Isolation Trees, the building blocks of the Isolation Forest. In this paper we propose effective, yet computationally inexpensive, methods to define feature importance scores at both global and local level for the Isolation Forest. Moreover, we define a procedure to perform unsupervised feature selection for Anomaly Detection problems based on our interpretability method; such procedure also serves the purpose of tackling the challenging task of feature importance evaluation in unsupervised anomaly detection. We assess the performance on several synthetic and real-world datasets, including comparisons against state-of-the-art interpretability techniques, and make the code publicly available to enhance reproducibility and foster research in the field.
Interpretable Machine Learning (IML) methods are used to gain insight into the relevance of a feature of interest for the performance of a model. Commonly used IML methods differ in whether they consider features of interest in isolation, e.g., Permutation Feature Importance (PFI), or in relation to all remaining feature variables, e.g., Conditional Feature Importance (CFI). As such, the perturbation mechanisms inherent to PFI and CFI represent extreme reference points. We introduce Relative Feature Importance (RFI), a generalization of PFI and CFI that allows for a more nuanced feature importance computation beyond the PFI versus CFI dichotomy. With RFI, the importance of a feature relative to any other subset of features can be assessed, including variables that were not available at training time. We derive general interpretation rules for RFI based on a detailed theoretical analysis of the implications of relative feature relevance, and demonstrate the method's usefulness on simulated examples.
An increasing number of model-agnostic interpretation techniques for machine learning (ML) models such as partial dependence plots (PDP), permutation feature importance (PFI) and Shapley values provide insightful model interpretations, but can lead to wrong conclusions if applied incorrectly. We highlight many general pitfalls of ML model interpretation, such as using interpretation techniques in the wrong context, interpreting models that do not generalize well, ignoring feature dependencies, interactions, uncertainty estimates and issues in high-dimensional settings, or making unjustified causal interpretations, and illustrate them with examples. We focus on pitfalls for global methods that describe the average model behavior, but many pitfalls also apply to local methods that explain individual predictions. Our paper addresses ML practitioners by raising awareness of pitfalls and identifying solutions for correct model interpretation, but also addresses ML researchers by discussing open issues for further research.
Graph Neural Networks (GNNs) are versatile, powerful machine learning methods that enable graph structure and feature representation learning, and have applications across many domains. For applications critically requiring interpretation, attention-based GNNs have been leveraged. However, these approaches either rely on specific model architectures or lack a joint consideration of graph structure and node features in their interpretation. Here we present a model-agnostic framework for interpreting important graph structure and node features, Graph neural networks Including SparSe inTerpretability (GISST). With any GNN model, GISST combines an attention mechanism and sparsity regularization to yield an important subgraph and node feature subset related to any graph-based task. Through a single self-attention layer, a GISST model learns an importance probability for each node feature and edge in the input graph. By including these importance probabilities in the model loss function, the probabilities are optimized end-to-end and tied to the task-specific performance. Furthermore, GISST sparsifies these importance probabilities with entropy and L1 regularization to reduce noise in the input graph topology and node features. Our GISST models achieve superior node feature and edge explanation precision in synthetic datasets, as compared to alternative interpretation approaches. Moreover, our GISST models are able to identify important graph structure in real-world datasets. We demonstrate in theory that edge feature importance and multiple edge types can be considered by incorporating them into the GISST edge probability computation. By jointly accounting for topology, node features, and edge features, GISST inherently provides simple and relevant interpretations for any GNN models and tasks.
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.
Human explanation (e.g., in terms of feature importance) has been recently used to extend the communication channel between human and agent in interactive machine learning. Under this setting, human trainers provide not only the ground truth but also some form of explanation. However, this kind of human guidance was only investigated in supervised learning tasks, and it remains unclear how to best incorporate this type of human knowledge into deep reinforcement learning. In this paper, we present the first study of using human visual explanations in human-in-the-loop reinforcement learning (HRL). We focus on the task of learning from feedback, in which the human trainer not only gives binary evaluative "good" or "bad" feedback for queried state-action pairs, but also provides a visual explanation by annotating relevant features in images. We propose EXPAND (EXPlanation AugmeNted feeDback) to encourage the model to encode task-relevant features through a context-aware data augmentation that only perturbs irrelevant features in human salient information. We choose five tasks, namely Pixel-Taxi and four Atari games, to evaluate the performance and sample efficiency of this approach. We show that our method significantly outperforms methods leveraging human explanation that are adapted from supervised learning, and Human-in-the-loop RL baselines that only utilize evaluative feedback.
In recent years, hyperparameter optimization (HPO) has become an increasingly important issue in the field of machine learning for the development of more accurate forecasting models. In this study, we explore the potential of HPO in modeling stock returns using a deep neural network (DNN). The potential of this approach was evaluated using technical indicators and fundamentals examined based on the effect the regularization of dropouts and batch normalization for all input data. We found that the model using technical indicators and dropout regularization significantly outperforms three other models, showing a positive predictability of 0.53% in-sample and 1.11% out-of-sample, thereby indicating the possibility of beating the historical average. We also demonstrate the stability of the model in terms of the changes in its feature importance over time.
The true population-level importance of a variable in a prediction task provides useful knowledge about the underlying data-generating mechanism and can help in deciding which measurements to collect in subsequent experiments. Valid statistical inference on this importance is a key component in understanding the population of interest. We present a computationally efficient procedure for estimating and obtaining valid statistical inference on the Shapley Population Variable Importance Measure (SPVIM). Although the computational complexity of the true SPVIM scales exponentially with the number of variables, we propose an estimator based on randomly sampling only $\Theta(n)$ feature subsets given $n$ observations. We prove that our estimator converges at an asymptotically optimal rate. Moreover, by deriving the asymptotic distribution of our estimator, we construct valid confidence intervals and hypothesis tests. Our procedure has good finite-sample performance in simulations, and for an in-hospital mortality prediction task produces similar variable importance estimates when different machine learning algorithms are applied.
Sub-seasonal climate forecasting (SSF) focuses on predicting key climate variables such as temperature and precipitation in the 2-week to 2-month time scales. Skillful SSF would have immense societal value, in areas such as agricultural productivity, water resource management, transportation and aviation systems, and emergency planning for extreme weather events. However, SSF is considered more challenging than either weather prediction or even seasonal prediction. In this paper, we carefully study a variety of machine learning (ML) approaches for SSF over the US mainland. While atmosphere-land-ocean couplings and the limited amount of good quality data makes it hard to apply black-box ML naively, we show that with carefully constructed feature representations, even linear regression models, e.g., Lasso, can be made to perform well. Among a broad suite of 10 ML approaches considered, gradient boosting performs the best, and deep learning (DL) methods show some promise with careful architecture choices. Overall, suitable ML methods are able to outperform the climatological baseline, i.e., predictions based on the 30-year average at a given location and time. Further, based on studying feature importance, ocean (especially indices based on climatic oscillations such as El Nino) and land (soil moisture) covariates are found to be predictive, whereas atmospheric covariates are not considered helpful.
Attention-based methods have played important roles in model interpretations, where the calculated attention weights are expected to highlight the critical parts of inputs~(e.g., keywords in sentences). However, recent research found that attention-as-importance interpretations often do not work as we expected. For example, learned attention weights sometimes highlight less meaningful tokens like "[SEP]", ",", and ".", and are frequently uncorrelated with other feature importance indicators like gradient-based measures. A recent debate over whether attention is an explanation or not has drawn considerable interest. In this paper, we demonstrate that one root cause of this phenomenon is the combinatorial shortcuts, which means that, in addition to the highlighted parts, the attention weights themselves may carry extra information that could be utilized by downstream models after attention layers. As a result, the attention weights are no longer pure importance indicators. We theoretically analyze combinatorial shortcuts, design one intuitive experiment to show their existence, and propose two methods to mitigate this issue. We conduct empirical studies on attention-based interpretation models. The results show that the proposed methods can effectively improve the interpretability of attention mechanisms.
Practitioners use feature importance to rank and eliminate weak predictors during model development in an effort to simplify models and improve generality. Unfortunately, they also routinely conflate such feature importance measures with feature impact, the isolated effect of an explanatory variable on the response variable. This can lead to real-world consequences when importance is inappropriately interpreted as impact for business or medical insight purposes. The dominant approach for computing importances is through interrogation of a fitted model, which works well for feature selection, but gives distorted measures of feature impact. The same method applied to the same data set can yield different feature importances, depending on the model, leading us to conclude that impact should be computed directly from the data. While there are nonparametric feature selection algorithms, they typically provide feature rankings, rather than measures of impact or importance. They also typically focus on single-variable associations with the response. In this paper, we give mathematical definitions of feature impact and importance, derived from partial dependence curves, that operate directly on the data. To assess quality, we show that features ranked by these definitions are competitive with existing feature selection techniques using three real data sets for predictive tasks.
The interpretation of feature importance in machine learning models is challenging when features are dependent. Permutation feature importance (PFI) ignores such dependencies, which can cause misleading interpretations due to extrapolation. A possible remedy is more advanced conditional PFI approaches that enable the assessment of feature importance conditional on all other features. Due to this shift in perspective and in order to enable correct interpretations, it is therefore important that the conditioning is transparent and humanly comprehensible. In this paper, we propose a new sampling mechanism for the conditional distribution based on permutations in conditional subgroups. As these subgroups are constructed using decision trees (transformation trees), the conditioning becomes inherently interpretable. This not only provides a simple and effective estimator of conditional PFI, but also local PFI estimates within the subgroups. In addition, we apply the conditional subgroups approach to partial dependence plots (PDP), a popular method for describing feature effects that can also suffer from extrapolation when features are dependent and interactions are present in the model. We show that PFI and PDP based on conditional subgroups often outperform methods such as conditional PFI based on knockoffs, or accumulated local effect plots. Furthermore, our approach allows for a more fine-grained interpretation of feature effects and importance within the conditional subgroups.
This paper introduces X-SHAP, a model-agnostic method that assesses multiplicative contributions of variables for both local and global predictions. This method theoretically and operationally extends the so-called additive SHAP approach. It proves useful underlying multiplicative interactions of factors, typically arising in sectors where Generalized Linear Models are traditionally used, such as in insurance or biology. We test the method on various datasets and propose a set of techniques based on individual X-SHAP contributions to build aggregated multiplicative contributions and to capture multiplicative feature importance, that we compare to traditional techniques.
Physicians taking care of patients with coronavirus disease (COVID-19) have described different changes in routine blood parameters. However, these changes, hinder them from performing COVID-19 diagnosis. We constructed a machine learning predictive model for COVID-19 diagnosis. The model was based and cross-validated on the routine blood tests of 5,333 patients with various bacterial and viral infections, and 160 COVID-19-positive patients. We selected operational ROC point at a sensitivity of 81.9% and specificity of 97.9%. The cross-validated area under the curve (AUC) was 0.97. The five most useful routine blood parameters for COVID19 diagnosis according to the feature importance scoring of the XGBoost algorithm were MCHC, eosinophil count, albumin, INR, and prothrombin activity percentage. tSNE visualization showed that the blood parameters of the patients with severe COVID-19 course are more like the parameters of bacterial than viral infection. The reported diagnostic accuracy is at least comparable and probably complementary to RT-PCR and chest CT studies. Patients with fever, cough, myalgia, and other symptoms can now have initial routine blood tests assessed by our diagnostic tool. All patients with a positive COVID-19 prediction would then undergo standard RT-PCR studies to confirm the diagnosis. We believe that our results present a significant contribution to improvements in COVID-19 diagnosis.
Globally increasing migration pressures call for new modelling approaches in order to design effective policies. It is important to have not only efficient models to predict migration flows but also to understand how specific parameters influence these flows. In this paper, we propose an artificial neural network (ANN) to model international migration. Moreover, we use a technique for interpreting machine learning models, namely Partial Dependence Plots (PDP), to show that one can well study the effects of drivers behind international migration. We train and evaluate the model on a dataset containing annual international bilateral migration from $1960$ to $2010$ from $175$ origin countries to $33$ mainly OECD destinations, along with the main determinants as identified in the migration literature. The experiments carried out confirm that: 1) the ANN model is more efficient w.r.t. a traditional model, and 2) using PDP we are able to gain additional insights on the specific effects of the migration drivers. This approach provides much more information than only using the feature importance information used in previous works.
With advances in reinforcement learning (RL), agents are now being developed in high-stakes application domains such as healthcare and transportation. Explaining the behavior of these agents is challenging, as the environments in which they act have large state spaces, and their decision-making can be affected by delayed rewards, making it difficult to analyze their behavior. To address this problem, several approaches have been developed. Some approaches attempt to convey the $\textit{global}$ behavior of the agent, describing the actions it takes in different states. Other approaches devised $\textit{local}$ explanations which provide information regarding the agent's decision-making in a particular state. In this paper, we combine global and local explanation methods, and evaluate their joint and separate contributions, providing (to the best of our knowledge) the first user study of combined local and global explanations for RL agents. Specifically, we augment strategy summaries that extract important trajectories of states from simulations of the agent with saliency maps which show what information the agent attends to. Our results show that the choice of what states to include in the summary (global information) strongly affects people's understanding of agents: participants shown summaries that included important states significantly outperformed participants who were presented with agent behavior in a randomly set of chosen world-states. We find mixed results with respect to augmenting demonstrations with saliency maps (local information), as the addition of saliency maps did not significantly improve performance in most cases. However, we do find some evidence that saliency maps can help users better understand what information the agent relies on in its decision making, suggesting avenues for future work that can further improve explanations of RL agents.