Research Papers

Induction of Non-monotonic Logic Programs To Explain Statistical Learning Models

Farhad Shakerin

We present a fast and scalable algorithm to induce non-monotonic logic programs from statistical learning models. We reduce the problem of search for best clauses to instances of the High-Utility Itemset Mining (HUIM) problem. In the HUIM problem, feature values and their importance are treated as transactions and utilities respectively. We make use of TreeExplainer, a fast and scalable implementation of the Explainable AI tool SHAP, to extract locally important features and their weights from ensemble tree models. Our experiments with UCI standard benchmarks suggest a significant improvement in terms of classification evaluation metrics and running time of the training algorithm compared to ALEPH, a state-of-the-art Inductive Logic Programming (ILP) system.

Algorithms now permeate multiple aspects of human lives and multiple recent results have reported that these algorithms may have biases pertaining to gender, race, and other demographic characteristics. The metrics used to quantify such biases have still focused on a static notion of algorithms. However, algorithms evolve over time. For instance, Tay (a conversational bot launched by Microsoft) was arguably not biased at its launch but quickly became biased, sexist, and racist over time. We suggest a set of intuitive metrics to study the variations in biases over time and present the results for a case study for genders represented in images resulting from a Twitter image search for #Nurse and #Doctor over a period of 21 days. Results indicate that biases vary significantly over time and the direction of bias could appear to be different on different days. Hence, one-shot measurements may not suffice for understanding algorithmic bias, thus motivating further work on studying biases in algorithms over time.

The impact of designing for security of AI is critical for humanity in the AI era. With humans increasingly becoming dependent upon AI, there is a need for neural networks that work reliably, inspite of Adversarial attacks. The vision for Safe and secure AI for popular use is achievable. To achieve safety of AI, this paper explores strategies and a novel deep learning architecture. To guard AI from adversaries, paper explores combination of 3 strategies: 1. Introduce randomness at inference time to hide the representation learning from adversaries. 2. Detect presence of adversaries by analyzing the sequence of inferences. 3. Exploit visual similarity. To realize these strategies, this paper designs a novel architecture, Dynamic Neural Defense, DND. This defense has 3 deep learning architectural features: 1. By hiding the way a neural network learns from exploratory attacks using a random computation graph, DND evades attack. 2. By analyzing input sequence to cloud AI inference engine with LSTM, DND detects attack sequence. 3. By inferring with visual similar inputs generated by VAE, any AI defended by DND approach does not succumb to hackers. Thus, a roadmap to develop reliable, safe and secure AI is presented.

The young field of AI Safety is still in the process of identifying its challenges and limitations. In this paper, we formally describe one such impossibility result, namely Unpredictability of AI. We prove that it is impossible to precisely and consistently predict what specific actions a smarter-than-human intelligent system will take to achieve its objectives, even if we know terminal goals of the system. In conclusion, impact of Unpredictability on AI Safety is discussed.

This work describes a large-scale analysis of sentiment associations in popular word embedding models along the lines of gender and ethnicity but also along the less frequently studied dimensions of socioeconomic status, age, sexual orientation, religious sentiment and political leanings. Consistent with previous scholarly literature, this work has found systemic bias against given names popular among African-Americans in most embedding models examined. Gender bias in embedding models however appears to be multifaceted and often reversed in polarity to what has been regularly reported. Interestingly, using the common operationalization of the term bias in the fairness literature, novel types of so far unreported bias types in word embedding models have also been identified. Specifically, the popular embedding models analyzed here display negative biases against middle and working-class socioeconomic status, male children, senior citizens, plain physical appearance, Islamic religious faith, non-religiosity and conservative political orientation. Reasons for the paradoxical underreporting of these bias types in the relevant literature are probably manifold but widely held blind spots when searching for algorithmic bias and a lack of widespread technical jargon to unambiguously describe a variety of algorithmic associations could conceivably be playing a role. The causal origins for the multiplicity of loaded associations attached to distinct demographic groups within embedding models are often unclear but the heterogeneity of said associations and their potential multifactorial roots raises doubts about the validity of grouping them all under the umbrella term bias. Richer and more fine-grained terminology as well as a more comprehensive exploration of the bias landscape could help the fairness epistemic community to characterize and neutralize algorithmic discrimination more efficiently.

Academic performance is perceived as a product of complex interactions between students' overall experience, personal characteristics and upbringing. Data science techniques, most commonly involving regression analysis and related approaches, serve as a viable means to explore this interplay. However, these tend to extract factors with wide-ranging impact, while overlooking variations specific to individual students. Focusing on each student's peculiarities is generally impossible with thousands or even hundreds of subjects, yet data mining methods might prove effective in devising more targeted approaches. For instance, subjects with shared characteristics can be assigned to clusters, which can then be examined separately with machine learning algorithms, thereby providing a more nuanced view of the factors affecting individuals in a particular group. In this context, we introduce a data science workflow allowing for fine-grained analysis of academic performance correlates that captures the subtle differences in students' sensitivities to these factors. Leveraging the Local Interpretable Model-Agnostic Explanations (LIME) algorithm from the toolbox of Explainable Artificial Intelligence (XAI) techniques, the proposed pipeline yields groups of students having similar academic attainment indicators, rather than similar features (e.g. familial background) as typically practiced in prior studies. As a proof-of-concept case study, a rich longitudinal dataset is selected to evaluate the effectiveness of the proposed approach versus a standard regression model.