In explainable machine learning, global feature importance methods try to determine how much each individual feature contributes to predicting the target variable, resulting in one importance score for each feature. But often, predicting the target variable requires interactions between several features (such as in the XOR function), and features might have complex statistical dependencies that allow to partially replace one feature with another one. In commonly used feature importance scores, these cooperative effects are conflated with the features’ individual contributions, making them prone to misinterpretations. In this work, we derive DIP, a new mathematical decomposition of individual feature importance scores that disentangles three components: the standalone contribution and the contributions stemming from interactions and dependencies. We prove that the DIP decomposition is unique and show how it can be estimated in practice. Based on these results, we propose a new visualization of feature importance scores that clearly illustrates the different contributions.

SHAP is one of the most popular \textit{local} feature-attribution methods. Given a function $f$ and an input $x \in \mathbb{R}^d$, it quantifies each feature’s contribution to $f(x)$. Recently, SHAP has been increasingly used for \textit{global} insights: practitioners average the absolute SHAP values over many data points to compute global feature importance scores, which are then used to discard “unimportant” features. % In this work, we investigate the soundness of this practice by asking whether small aggregate SHAP values necessarily imply that the corresponding feature does not affect the function. Unfortunately, the answer is no: even if the $i$-th SHAP value equals $0$ on the entire data support, there exist functions that clearly depend on Feature $i$. The issue is that computing SHAP values involves evaluating $f$ on points outside of the data support, where $f$ can be strategically designed to mask its dependence on Feature $i$. % To address this, we propose to aggregate SHAP values over the \textit{extended} support, which is the product of the marginals of the underlying distribution. With this modification, we show that a small aggregate SHAP value implies that we can safely discard the corresponding feature. % We then extend our results to KernelSHAP, the most popular method to approximate SHAP values in practice. We show that if KernelSHAP is computed over the extended distribution, a small aggregate KernelSHAP value justifies feature removal. This result holds independently of whether KernelSHAP accurately approximates true SHAP values, making it one of the first theoretical results to characterize the KernelSHAP algorithm itself. Our findings have both theoretical and practical implications. We introduce the "Shapley Lie algebra", which offers algebraic insights that may enable a deeper investigation of SHAP and we show that a simple preprocessing step – randomly permuting each column of the data matrix – enables safely discarding features based on aggregate SHAP and KernelSHAP values.

The leakage of benchmark data into the training data has emerged as a significant challenge for evaluating the capabilities of large language models (LLMs). In this work, we challenge the common assumption that small-scale contamination renders benchmark evaluations invalid. First, we experimentally quantify the magnitude of benchmark overfitting based on scaling along three dimensions: The number of model parameters (up to 1.6B), the number of times an example is seen (up to 144), and the number of training tokens (up to 40B). If model and data follow the Chinchilla scaling laws, minor contamination indeed leads to overfitting. At the same time, even 144 times of contamination can be forgotten if the training data is scaled beyond five times Chinchilla, a regime characteristic of many modern LLMs. Continual pre-training of OLMo-7B corroborates these results. Next, we study the impact of the weight decay parameter on example forgetting, showing that empirical forgetting occurs faster than the cumulative weight decay. This allows us to gauge the degree of example forgetting in large-scale training runs, indicating that many LLMs, including Llama 3 405B, have forgotten the data seen at the beginning of training.

Scaling limits, such as infinite-width limits, serve as promising theoretical tools to study large-scale models. However, it is widely believed that existing infinite-width theory does not faithfully explain the behavior of practical networks, especially those trained in *standard parameterization* (SP) meaning He initialization with a global learning rate. For instance, existing theory for SP predicts instability at large learning rates and vanishing feature learning at stable ones. In practice, however, optimal learning rates decay slower than theoretically predicted and networks exhibit both stable training and non-trivial feature learning, even at very large widths. Here, we show that this discrepancy is not fully explained by finite-width phenomena. Instead, we find a resolution through a finer-grained analysis of the regime previously considered unstable and therefore uninteresting. In particular, we show that, under the cross-entropy (CE) loss, the unstable regime comprises two distinct sub-regimes: a catastrophically unstable regime and a more benign controlled divergence regime, where logits diverge but gradients and activations remain stable. Moreover, under large learning rates at the edge of the controlled divergence regime, there exists a well-defined infinite width limit where features continue to evolve in all the hidden layers. In experiments across optimizers, architectures, and data modalities, we validate that neural networks operate in this controlled divergence regime under CE loss but not under MSE loss. Our empirical evidence suggests that width-scaling considerations are surprisingly useful for predicting empirically maximal stable learning rate exponents which provide useful guidance on optimal learning rate exponents. Finally, our analysis clarifies the effectiveness and limitations of recently proposed layerwise learning rate scalings for standard initialization.

When applicants get rejected by a high-stakes algorithmic decision system, recourse explanations provide actionable suggestions for applicants on how to change their input features to get a positive evaluation. A crucial yet overlooked phenomenon is that recourse explanations are *performative*: When many applicants act according to their recommendations, their collective behavior may shift the data distribution and, once the model is refitted, also the decision boundary. Consequently, the recourse algorithm may render its own recommendations *invalid*, such that applicants who make the effort of implementing their recommendations may be rejected again when they reapply. In this work, we formally characterize the conditions under which recourse explanations remain valid under their own performative effects. In particular, we prove that recourse actions may become invalid if they are influenced by or if they intervene on non-causal variables. Based on this analysis, we caution against the use of standard counterfactual explanations and causal recourse methods, and instead advocate for recourse methods that recommend actions exclusively on causal variables.

In sensitive contexts, providers of machine learning algorithms are increasingly required to give explanations for their algorithms' decisions. However, explanation receivers might not trust the provider, who potentially could output misleading or manipulated explanations. In this work, we investigate an auditing framework in which a third-party auditor or a collective of users attempts to sanity-check explanations: they can query model decisions and the corresponding local explanations, pool all the information received, and then check for basic consistency properties. We prove upper and lower bounds on the amount of queries that are needed for an auditor to succeed within this framework. Our results show that successful auditing requires a potentially exorbitant number of queries -- particularly in high dimensional cases. Our analysis also reveals that a key property is the ``locality'' of the provided explanations --- a quantity that so far has not been paid much attention to in the explainability literature. Looking forward, our results suggest that for complex high-dimensional settings, merely providing a pointwise prediction and explanation could be insufficient, as there is no way for the users to verify that the provided explanations are not completely made-up.

In explainable machine learning, local post-hoc explanation algorithms and inherently interpretable models are often seen as competing approaches. This work offers a partial reconciliation between the two by establishing a correspondence between Shapley Values and Generalized Additive Models (GAMs). We introduce $n$-Shapley Values, a parametric family of local post-hoc explanation algorithms that explain individual predictions with interaction terms up to order $n$. By varying the parameter $n$, we obtain a sequence of explanations that covers the entire range from Shapley Values up to a uniquely determined decomposition of the function we want to explain. The relationship between $n$-Shapley Values and this decomposition offers a functionally-grounded characterization of Shapley Values, which highlights their limitations. We then show that $n$-Shapley Values, as well as the Shapley Taylor- and Faith-Shap interaction indices, recover GAMs with interaction terms up to order $n$. This implies that the original Shapely Values recover GAMs without variable interactions. Taken together, our results provide a precise characterization of Shapley Values as they are being used in explainable machine learning. They also offer a principled interpretation of partial dependence plots of Shapley Values in terms of the underlying functional decomposition. A package for the estimation of different interaction indices is available at https://github.com/tml-tuebingen/nshap.

The success of over-parameterized neural networks trained to near-zero training error has caused great interest in the phenomenon of benign overfitting, where estimators are statistically consistent even though they interpolate noisy training data. While benign overfitting in fixed dimension has been established for some learning methods, current literature suggests that for regression with typical kernel methods and wide neural networks, benign overfitting requires a high-dimensional setting, where the dimension grows with the sample size. In this paper, we show that the smoothness of the estimators, and not the dimension, is the key: benign overfitting is possible if and only if the estimator's derivatives are large enough. We generalize existing inconsistency results to non-interpolating models and more kernels to show that benign overfitting with moderate derivatives is impossible in fixed dimension. Conversely, we show that benign overfitting is possible for regression with a sequence of spiky-smooth kernels with large derivatives. Using neural tangent kernels, we translate our results to wide neural networks. We prove that while infinite-width networks do not overfit benignly with the ReLU activation, this can be fixed by adding small high-frequency fluctuations to the activation function. Our experiments verify that such neural networks, while overfitting, can indeed generalize well even on low-dimensional data sets.

Originally, tangles were invented as an abstract tool in mathematical graph theory to prove the famous graph minor theorem. In this paper, we showcase the practical potential of tangles in machine learning applications. Given a collection of cuts of any dataset, tangles aggregate these cuts to point in the direction of a dense structure. As a result, a cluster is softly characterized by a set of consistent pointers. This highly flexible approach can solve clustering problems in various setups, ranging from questionnaires over community detection in graphs to clustering points in metric spaces. The output of our proposed framework is hierarchical and induces the notion of a soft dendrogram, which can help explore the cluster structure of a dataset. The computational complexity of aggregating the cuts is linear in the number of data points. Thus the bottleneck of the tangle approach is to generate the cuts, for which simple and fast algorithms form a sufficient basis. In our paper we construct the algorithmic framework for clustering with tangles, prove theoretical guarantees in various settings, and provide extensive simulations and use cases. Python code is available on github.

The objective of ordinal embedding is to find a Euclidean representation of a set of abstract items, using only answers to triplet comparisons of the form “Is item $i$ closer to item $j$ or item $k$?”. In recent years, numerous algorithms have been proposed to solve this problem. However, there does not exist a fair and thorough assessment of these embedding methods and therefore several key questions remain unanswered: Which algorithms perform better when the embedding dimension is constrained or few triplet comparisons are available? Which ones scale better with increasing sample size or dimension? In our paper, we address these questions and provide an extensive and systematic empirical evaluation of existing algorithms as well as a new neural network approach. We find that simple, relatively unknown, non-convex methods consistently outperform all other algorithms across a broad range of tasks including more recent and elaborate methods based on neural networks or landmark approaches. This finding can be explained by the insight that many of the non-convex optimization approaches do not suffer from local optima. Our comprehensive assessment is enabled by our unified library of popular embedding algorithms that leverages GPU resources and allows for fast and accurate embeddings of millions of data points.

Applications of machine learning inform human decision makers in a broad range of tasks. The resulting problem is usually formulated in terms of a single decision maker. We argue that it should rather be described as a two-player learning problem where one player is the machine and the other the human. While both players try to optimize the final decision, the setup is often characterized by (1) the presence of private information and (2) opacity, that is imperfect understanding between the decision makers. We prove that both properties can complicate decision making considerably. A lower bound quantifies the worst-case hardness of optimally advising a decision maker who is opaque or has access to private information. An upper bound shows that a simple coordination strategy is nearly minimax optimal. More efficient learning is possible under certain assumptions on the problem, for example that both players learn to take actions independently. Such assumptions are implicit in existing literature, for example in medical applications of machine learning, but have not been described or justified theoretically.

Full Bayesian posteriors are rarely analytically tractable, which is why real-world Bayesian inference heavily relies on approximate techniques. Approximations generally differ from the true posterior and require diagnostic tools to assess whether the inference can still be trusted. We investigate a new approach to diagnosing approximate inference: the approximation mismatch is attributed to a change in the inductive bias by treating the approximations as exact and reverse-engineering the corresponding prior. We show that the problem is more complicated than it appears to be at first glance, because the solution generally depends on the observation. By reframing the problem in terms of incompatible conditional distributions we arrive at a natural solution: the Gibbs prior. The resulting diagnostic is based on pseudo-Gibbs sampling, which is widely applicable and easy to implement. We illustrate how the Gibbs prior can be used to discover the inductive bias in a controlled Gaussian setting and for a variety of Bayesian models and approximations.

Recent work shows that in complex model classes, interpolators can achieve statistical generalization and even be optimal for statistical learning. However, despite increasing interest in learning models with good causal properties, there is no understanding of whether such interpolators can also achieve causal generalization. To address this gap, we study causal learning from observational data through the lens of interpolation and its counterpart---regularization. Under a simple linear causal model, we derive precise asymptotics for the causal risk of the min-norm interpolator and ridge regressors in the high-dimensional regime. We find a large range of behavior that can be precisely characterized by a new measure of confounding strength. When confounding strength is positive, which holds under independent causal mechanisms---a standard assumption in causal learning---we find that interpolators cannot be optimal. Indeed, causal learning requires stronger regularization than statistical learning. Beyond this assumption, when confounding is negative, we observe a phenomenon of self-induced regularization due to positive alignment between statistical and causal signals. Here, causal learning requires weaker regularization than statistical learning, interpolators can be optimal, and optimal regularization can even be negative.

Despite the ubiquity of kernel-based clustering, surprisingly few statistical guarantees exist beyond settings that consider strong structural assumptions on the data generation process. In this work, we take a step towards bridging this gap by studying the statistical performance of kernel-based clustering algorithms under non-parametric mixture models. We provide necessary and sufficient separability conditions under which these algorithms can consistently recover the underlying true clustering. Our analysis provides guarantees for kernel clustering approaches without structural assumptions on the form of the component distributions. Additionally, we establish a key equivalence between kernel-based data-clustering and kernel density-based clustering. This enables us to provide consistency guarantees for kernel-based estimators of non-parametric mixture models. Along with theoretical implications, this connection could have practical implications, including in the systematic choice of the bandwidth of the Gaussian kernel in the context of clustering.

Machine learning is used more and more often for sensitive applications, sometimes replacing humans in critical decision-making processes. As such, interpretability of these algorithms is a pressing need. One popular algorithm to provide interpretability is LIME (Local Interpretable Model-Agnostic Explanation). In this paper, we provide the first theoretical analysis of LIME. We derive closed-form expressions for the coefficients of the interpretable model when the function to explain is linear. The good news is that these coefficients are proportional to the gradient of the function to explain: LIME indeed discovers meaningful features. However, our analysis also reveals that poor choices of parameters can lead LIME to miss important features.

We address the problem of classification under fairness constraints. Given a notion of fairness, the goal is to learn a classifier that is not discriminatory against a group of individuals. In the literature, this problem is often formulated as a constrained optimization problem and solved using relaxations of the fairness constraints. We show that many existing relaxations are unsatisfactory: even if a model satisfies the relaxed constraint, it can be surprisingly unfair. We propose a principled framework to solve this problem. This new approach uses a strongly convex formulation and comes with theoretical guarantees on the fairness of its solution. In practice, we show that this method gives promising results on real data.

A graph generative model takes a graph as input and is supposed to generate new graphs that “look like” the input graph. While most classical models focus on few, hand-selected graph statistics and are too simplistic to reproduce real-world graphs, NetGAN recently emerged as an attractive alternative: by training a GAN to learn the random walk distribution of the input graph, the algorithm is able to reproduce a large number of important network patterns simultaneously, without explicitly specifying any of them. In this paper, we investigate the implicit bias of NetGAN. We find that the root of its generalization properties does not lie in the GAN architecture, but in an inconspicuous low-rank approximation of the logits random walk transition matrix. Step by step we can strip NetGAN of all unnecessary parts, including the GAN, and obtain a highly simplified reformulation that achieves comparable generalization results, but is orders of magnitudes faster and easier to adapt. Being much simpler on the conceptual side, we reveal the implicit inductive bias of the algorithm — an important step towards increasing the interpretability, transparency and acceptance of machine learning systems.

We consider the problem of classification in a comparison-based setting: given a set of objects, we only have access to triplet comparisons of the form ``object A is closer to object B than to object C.'' In this paper we introduce TripletBoost, a new method that can learn a classifier just from such triplet comparisons. The main idea is to aggregate the triplets information into weak classifiers, which can subsequently be boosted to a strong classifier. Our method has two main advantages: (i) it is applicable to data from any metric space, and (ii) it can deal with large scale problems using only passively obtained and noisy triplets. We derive theoretical generalization guarantees and a lower bound on the number of necessary triplets, and we empirically show that our method is both competitive with state of the art approaches and resistant to noise.

We address the classical problem of hierarchical clustering, but in a framework where one does not have access to a representation of the objects or their pairwise similarities. Instead, we assume that only a set of comparisons between objects is available, that is, statements of the form ``objects i and j are more similar than objects k and l.'' Such a scenario is commonly encountered in crowdsourcing applications. The focus of this work is to develop comparison-based hierarchical clustering algorithms that do not rely on the principles of ordinal embedding. We show that single and complete linkage are inherently comparison-based and we develop variants of average linkage. We provide statistical guarantees for the different methods under a planted hierarchical partition model. We also empirically demonstrate the performance of the proposed approaches on several datasets.

Assume we are given a set of items from a general metric space, but we neither have access to the representation of the data nor to the distances between data points. Instead, suppose that we can actively choose a triplet of items (A, B, C) and ask an oracle whether item A is closer to item B or to item C. In this paper, we propose a novel random forest algorithm for regression and classification that relies only on such triplet comparisons. In the theory part of this paper, we establish sufficient conditions for the consistency of such a forest. In a set of comprehensive experiments, we then demonstrate that the proposed random forest is efficient both for classification and regression. In particular, it is even competitive with other methods that have direct access to the metric representation of the data.

Hypothesis testing for graphs has been an important tool in applied research fields for more than two decades, and still remains a challenging problem as one often needs to draw inference from few replicates of large graphs. Recent studies in statistics and learning theory have provided some theoretical insights about such high-dimensional graph testing problems, but the practicality of the developed theoretical methods remains an open question. In this paper, we consider the problem of two-sample testing of large graphs. We demonstrate the practical merits and limitations of existing theoretical tests and their bootstrapped variants. We also propose two new tests based on asymptotic distributions. We show that these tests are computationally less expensive and, in some cases, more reliable than the existing methods.

Random forests are learning algorithms that build large collections of random trees and make predictions by averaging the individual tree predictions. In this paper, we consider various tree constructions and examine how the choice of parameters affects the generalization error of the resulting random forests as the sample size goes to infinity. We show that subsampling of data points during the tree construction phase is important: Forests can become inconsistent with either no subsampling or too severe subsampling. As a consequence, even highly randomized trees can lead to inconsistent forests if no subsampling is used, which implies that some of the commonly used setups for random forests can be inconsistent. As a second consequence we can show that trees that have good performance in nearest-neighbor search can be a poor choice for random forests.

Given data from a general metric space, one of the standard machine learning pipelines is to first embed the data into a Euclidean space and subsequently apply out of the box machine learning algorithms to analyze the data. The quality of such an embedding is typically described in terms of a distortion measure. In this paper, we show that many of the existing distortion measures behave in an undesired way, when considered from a machine learning point of view. We investigate desirable properties of distortion measures and formally prove that most of the existing measures fail to satisfy these properties. These theoretical findings are supported by simulations, which for example demonstrate that existing distortion measures are not robust to noise or outliers and cannot serve as good indicators for classification accuracy. As an alternative, we suggest a new measure of distortion, called $\sigma$-distortion. We can show both in theory and in experiments that it satisfies all desirable properties and is a better candidate to evaluate distortion in the context of machine learning.

Neural Information Processing Systems (NIPS) is a top-tier annual conference in machine learning. The 2016 edition of the conference comprised more than 2,400 paper submissions, 3,000 reviewers, and 8,000 attendees. This represents a growth of nearly 40% in terms of submissions, 96% in terms of reviewers, and over 100% in terms of attendees as compared to the previous year. The massive scale as well as rapid growth of the conference calls for a thorough quality assessment of the peer-review process and novel means of improvement. In this paper, we analyze several aspects of the data collected during the review process, including an experiment investigating the efficacy of collecting ordinal rankings from reviewers. We make a number of key observations, provide suggestions that may be useful for subsequent conferences, and discuss open problems towards the goal of improving peer review.

We consider machine learning in a comparison-based setting where we are given a set of points in a metric space, but we have no access to the actual distances between the points. Instead, we can only ask an oracle whether the distance between two points i and j is smaller than the distance between the points i and k. We are concerned with data structures and algorithms to find nearest neighbors based on such comparisons. We focus on a simple yet effective algorithm that recursively splits the space by first selecting two random pivot points and then assigning all other points to the closer of the two (comparison tree). We prove that if the metric space satisfies certain expansion conditions, then with high probability the height of the comparison tree is logarithmic in the number of points, leading to efficient search performance. We also provide an upper bound for the failure probability to return the true nearest neighbor. Experiments show that the comparison tree is competitive with algorithms that have access to the actual distance values, and needs less triplet comparisons than other competitors.

We consider a two-sample hypothesis testing problem, where the distributions are defined on the space of undirected graphs, and one has access to only one observation from each model. A motivating example for this problem is comparing the friendship networks on Facebook and LinkedIn. The practical approach to such problems is to compare the networks based on certain network statistics. In this paper, we present a general principle for two-sample hypothesis testing in such scenarios without making any assumption about the network generation process. The main contribution of the paper is a general formulation of the problem based on concentration of network statistics, and consequently, a consistent two-sample test that arises as the natural solution for this problem. We also show that the proposed test is minimax optimal for certain network statistics.

Given only information in the form of similarity triplets "Object A is more similar to object B than to object C" about a data set, we propose two ways of defining a kernel function on the data set. While previous approaches construct a low-dimensional Euclidean embedding of the data set that reflects the given similarity triplets, we aim at defining kernel functions that correspond to high-dimensional embeddings. These kernel functions can subsequently be used to apply any kernel method to the data set.

In recent years it has become popular to study machine learning problems in a setting of ordinal distance information rather than numerical distance measurements. By ordinal distance information we refer to binary answers to distance comparisons such as $d(A,B)<d(C,D)$. For many problems in machine learning and statistics it is unclear how to solve them in such a scenario. Up to now, the main approach is to explicitly construct an ordinal embedding of the data points in the Euclidean space, an approach that has a number of drawbacks. In this paper, we propose algorithms for the problems of medoid estimation, outlier identification, classification, and clustering when given only ordinal data. They are based on estimating the lens depth function and the $k$-relative neighborhood graph on a data set. Our algorithms are simple, are much faster than an ordinal embedding approach and avoid some of its drawbacks, and can easily be parallelized.

Peer grading is the process of students reviewing each others' work, such as homework submissions, and has lately become a popular mechanism used in massive open online courses (MOOCs). Intrigued by this idea, we used it in a course on algorithms and data structures at the University of Hamburg. Throughout the whole semester, students repeatedly handed in submissions to exercises, which were then evaluated both by teaching assistants and by a peer grading mechanism, yielding a large dataset of teacher and peer grades. We applied different statistical and machine learning methods to aggregate the peer grades in order to come up with accurate final grades for the submissions (supervised and unsupervised, methods based on numeric scores and ordinal rankings). Surprisingly, none of them improves over the baseline of using the mean peer grade as the final grade. We discuss a number of possible explanations for these results and present a thorough analysis of the generated dataset.

While existing methods for estimating the intrinsic dimension of datasets require to know distances between data points, we consider a situation where one has considerably less information. Given a sample of points, all we get to see is who are the k nearest neighbors of every point. In other words, we get the adjacency matrix of the directed, unweighted k-nearest neighbor graph on the sample, but do not know any point coordinates or distances between the points. We provide two estimators for this situation, a naive one and a more elaborate one. Both of them can be proved to be statistically consistent. However, further theoretical and experimental evidence shows that the naive estimator performs rather poorly, whereas the elaborate one achieves results comparable to those of estimators based on distance information.

We consider the problem of vector quantization of i.i.d. samples drawn from a density p on \mathbbR^d. It is desirable that the representatives selected by the quantization algorithm have the same distribution p as the original sample points. However, quantization algorithms based on Euclidean distance, such as k-means, do not have this property. We provide a solution to this problem that takes the unweighted k-nearest neighbor graph on the sample as input. In particular, it does not need to have access to the data points themselves. Our solution generates quantization centers that are “evenly spaced". We exploit this property to downsample geometric graphs and show that our method produces sparse downsampled graphs. Our algorithm is easy to implement, and we provide theoretical guarantees on the performance of the proposed algorithm.

Ordinal embedding refers to the following problem: all we know about an unknown set of points x_1,\ldots, x_n ∈\mathbbR^d are ordinal constraints of the form \|x_i - x_j\| < \|x_k - x_l\|; the task is to construct a realization y_1,\ldots, y_n ∈\mathbbR^d that preserves these ordinal constraints. It has been conjectured since the 1960ies that upon knowledge of all ordinal constraints a large but finite set of points can be approximately reconstructed up to a similarity transformation. The main result of our paper is a formal proof of this conjecture.

Consider a neighborhood graph, for example a k-nearest neighbor graph, that is constructed on sample points drawn according to some density p. Our goal is to re-weight the graph’s edges such that all cuts and volumes behave as if the graph was built on a different sample drawn from an alternative density q. We introduce the f-adjusted graph and prove that it provides the correct cuts and volumes as the sample size tends to infinity. From an algebraic perspective, we show that its normalized Laplacian, denoted as the f-adjusted Laplacian, represents a natural family of diagonal perturbations of the original normalized Laplacian. Our technique allows to apply any cut and volume based algorithm to the f-adjusted graph, for example spectral clustering, in order to study the given graph as if it were built on an unaccessible sample from a different density. We point out applications in sample bias correction, data uniformization, and multi-scale analysis of graphs.

We study the problem of ordinal embedding: given a set of ordinal constraints of the form distance(i,j) < distance(k,l) for some_quadruples (i,j,k,l) of indices, the goal is to construct a point configuration \hat\bmx_1, ..., \hat\bmx_n in \R^p that preserves these constraints as well as possible. Our first contribution is to suggest a simple new algorithm for this problem, Soft Ordinal Embedding. The key feature of the algorithm is that it recovers not only the ordinal constraints, but even the density structure of the underlying data set. As our second contribution we prove that in the large sample limit it is enough to know “local ordinal information” in order to perfectly reconstruct a given point configuration. This leads to our Local Ordinal Embedding algorithm, which can also be used for graph drawing.

For a density f on Rd, a high-density cluster is any connected component of {x : f (x) ≥ λ}, for some λ > 0. The set of all high-density clusters forms a hierarchy called the cluster tree of f . We present two procedures for estimating the cluster tree given samples from f . The first is a robust variant of the single linkage algorithm for hierarchical clustering. The second is based on the k-nearest neighbor graph of the samples. We give finite-sample convergence rates for these algorithms, which also imply consistency, and we derive lower bounds on the sample complexity of cluster tree estimation. Finally, we study a tree pruning procedure that guarantees, under milder conditions than usual, to remove clusters that are spurious while recovering those that are salient.

Consider an unweighted k-nearest neighbor graph on n points that have been sampled i.i.d. from some unknown density p on R^d. We prove how one can estimate the density p just from the unweighted adjacency matrix of the graph, without knowing the points themselves or their distance or similarity scores. The key insights are that local differences in link numbers can be used to estimate some local function of p, and that integrating this function along shortest paths leads to an estimate of the underlying density.

We study the family of p-resistances on graphs for p ≥ 1. This family generalizes the standard resistance distance. We prove that for any fixed graph, for p=1, the p-resistance coincides with the shortest path distance, for p=2 it coincides with the standard resistance distance, and for p → ∞ it converges to the inverse of the minimal s-t-cut in the graph. Secondly, we consider the special case of random geometric graphs (such as k-nearest neighbor graphs) when the number n of vertices in the graph tends to infinity. We prove that an interesting phase-transition takes place. There exist two critical thresholds p^* and p^** such that if p < p^, then the p-resistance depends on meaningful global properties of the graph, whereas if p > p^, it only depends on trivial local quantities and does not convey any useful information. We can explicitly compute the critical values: p^ = 1 + 1/(d-1) and p^** = 1 + 1/(d-2) where d is the dimension of the underlying space (we believe that the fact that there is a small gap between p^* and p^** is an artifact of our proofs. We also relate our findings to Laplacian regularization and suggest to use q-Laplacians as regularizers, where q satisfies 1/p^* + 1/q = 1.

We consider the problem of local graph clustering where the aim is to discover the local cluster corresponding to a point of interest. The most popular algorithms to solve this problem start a random walk at the point of interest and let it run until some stopping criterion is met. The vertices visited are then considered the local cluster. We suggest a more powerful alternative, the multi-agent random walk. It consists of several ``agents'' connected by a fixed rope of length l. All agents move independently like a standard random walk on the graph, but they are constrained to have distance at most l from each other. The main insight is that for several agents it is harder to simultaneously travel over the bottleneck of a graph than for just one agent. Hence, the multi-agent random walk has less tendency to mistakenly merge two different clusters than the original random walk. In our paper we analyze the multi-agent random walk theoretically and compare it experimentally to the major local graph clustering algorithms from the literature. We find that our multi-agent random walk consistently outperforms these algorithms.

The commute distance between two vertices in a graph is the expected time it takes a random walk to travel from the first to the second vertex and back. We study the behavior of the commute distance as the size of the underlying graph increases. We prove that the commute distance converges to an expression that does not take into account the structure of the graph at all and that is completely meaningless as a distance function on the graph. Consequently, the use of the raw commute distance for machine learning purposes is strongly discouraged for large graphs and in high dimensions. As an alternative we introduce the amplified commute distance that corrects for the undesired large sample effects.

Graph clustering methods such as spectral clustering are defined for general weighted graphs. In machine learning, however, data often is not given in form of a graph, but in terms of similarity (or distance) values between points. In this case, first a neighborhood graph is constructed using the similarities between the points and then a graph clustering algorithm is applied to this graph. In this paper we investigate the influence of the construction of the similarity graph on the clustering results. We first study the convergence of graph clustering criteria such as the normalized cut (Ncut) as the sample size tends to infinity. We find that the limit expressions are different for different types of graph, for example the r-neighborhood graph or the k -nearest neighbor graph. In plain words: Ncut on a kNN graph does something systematically different than Ncut on an r-neighborhood graph! This finding shows that graph clustering criteria cannot be studied independently of the kind of graph they are applied to. We also provide examples which show that these differences can be observed for toy and real data already for rather small sample sizes.

Clustering is often formulated as a discrete optimization problem. The objective is to find, among all partitions of the data set, the best one according to some quality measure. However, in the statistical setting where we assume that the finite data set has been sampled from some underlying space, the goal is not to find the best partition of the given sample, but to approximate the true partition of the under- lying space. We argue that the discrete optimization approach usually does not achieve this goal. As an alternative, we suggest the paradigm of “nearest neighbor clustering”. Instead of selecting the best out of all partitions of the sample, it only considers partitions in some restricted function class. Using tools from statistical learning theory we prove that nearest neighbor clustering is statistically consis- tent. Moreover, its worst case complexity is polynomial by construction, and it can be implemented with small average case complexity using branch and bound.

An important aspect of clustering algorithms is whether the partitions constructed on finite samples converge to a useful clustering of the whole data space as the sample size increases. This paper investigates this question for normalized and unnormalized versions of the popular spec- tral clustering algorithm. Surprisingly, the convergence of unnormalized spectral clustering is more difficult to handle than the normalized case. Even though recently some first results on the convergence of normal- ized spectral clustering have been obtained, for the unnormalized case we have to develop a completely new approach combining tools from numerical integration, spectral and perturbation theory, and probability. It turns out that while in the normalized case, spectral clustering usually converges to a nice partition of the data space, in the unnormalized case the same only holds under strong additional assumptions which are not always satisfied. We conclude that our analysis gives strong evidence for the superiority of normalized spectral clustering. It also provides a basis for future exploration of other Laplacian-based methods.