skip to main content

This content will become publicly available on June 30, 2023

Title: An Evaluative Measure of Clustering Methods Incorporating Hyperparameter Sensitivity
Clustering algorithms are often evaluated using metrics which compare with ground-truth cluster assignments, such as Rand index and NMI. Algorithm performance may vary widely for different hyperparameters, however, and thus model selection based on optimal performance for these metrics is discordant with how these algorithms are applied in practice, where labels are unavailable and tuning is often more art than science. It is therefore desirable to compare clustering algorithms not only on their optimally tuned performance, but also some notion of how realistic it would be to obtain this performance in practice. We propose an evaluation of clustering methods capturing this ease-of-tuning by modeling the expected best clustering score under a given computation budget. To encourage the adoption of the proposed metric alongside classic clustering evaluations, we provide an extensible benchmarking framework. We perform an extensive empirical evaluation of our proposed metric on popular clustering algorithms over a large collection of datasets from different domains, and observe that our new metric leads to several noteworthy observations.
Authors:
; ; ; ;
Award ID(s):
1763618
Publication Date:
NSF-PAR ID:
10356094
Journal Name:
Proceedings of the AAAI Conference on Artificial Intelligence
Volume:
36
Issue:
7
Page Range or eLocation-ID:
7788 to 7796
ISSN:
2159-5399
Sponsoring Org:
National Science Foundation
More Like this
  1. Abstract Statistical relational learning (SRL) frameworks are effective at defining probabilistic models over complex relational data. They often use weighted first-order logical rules where the weights of the rules govern probabilistic interactions and are usually learned from data. Existing weight learning approaches typically attempt to learn a set of weights that maximizes some function of data likelihood; however, this does not always translate to optimal performance on a desired domain metric, such as accuracy or F1 score. In this paper, we introduce a taxonomy of search-based weight learning approaches for SRL frameworks that directly optimize weights on a chosen domain performance metric. To effectively apply these search-based approaches, we introduce a novel projection, referred to as scaled space (SS), that is an accurate representation of the true weight space. We show that SS removes redundancies in the weight space and captures the semantic distance between the possible weight configurations. In order to improve the efficiency of search, we also introduce an approximation of SS which simplifies the process of sampling weight configurations. We demonstrate these approaches on two state-of-the-art SRL frameworks: Markov logic networks and probabilistic soft logic. We perform empirical evaluation on five real-world datasets and evaluate them eachmore »on two different metrics. We also compare them against four other weight learning approaches. Our experimental results show that our proposed search-based approaches outperform likelihood-based approaches and yield up to a 10% improvement across a variety of performance metrics. Further, we perform an extensive evaluation to measure the robustness of our approach to different initializations and hyperparameters. The results indicate that our approach is both accurate and robust.« less
  2. Recommendation and ranking systems are known to suffer from popularity bias; the tendency of the algorithm to favor a few popular items while under-representing the majority of other items. Prior research has examined various approaches for mitigating popularity bias and enhancing the recommendation of long-tail, less popular, items. The effectiveness of these approaches is often assessed using different metrics to evaluate the extent to which over-concentration on popular items is reduced. However, not much attention has been given to the user-centered evaluation of this bias; how different users with different levels of interest towards popular items are affected by such algorithms. In this paper, we show the limitations of the existing metrics to evaluate popularity bias mitigation when we want to assess these algorithms from the users’ perspective and we propose a new metric that can address these limitations. In addition, we present an effective approach that mitigates popularity bias from the user-centered point of view. Finally, we investigate several state-of-the-art approaches proposed in recent years to mitigate popularity bias and evaluate their performances using the existing metrics and also from the users’ perspective. Our experimental results using two publicly-available datasets show that existing popularity bias mitigation techniques ignore themore »users’ tolerance towards popular items. Our proposed user-centered method can tackle popularity bias effectively for different users while also improving the existing metrics.« less
  3. Abstract

    A common way to integrate and analyze large amounts of biological “omic” data is through pathway reconstruction: using condition-specific omic data to create a subnetwork of a generic background network that represents some process or cellular state. A challenge in pathway reconstruction is that adjusting pathway reconstruction algorithms’ parameters produces pathways with drastically different topological properties and biological interpretations. Due to the exploratory nature of pathway reconstruction, there is no ground truth for direct evaluation, so parameter tuning methods typically used in statistics and machine learning are inapplicable. We developed the pathway parameter advising algorithm to tune pathway reconstruction algorithms to minimize biologically implausible predictions. We leverage background knowledge in pathway databases to select pathways whose high-level structure resembles that of manually curated biological pathways. At the core of this method is a graphlet decomposition metric, which measures topological similarity to curated biological pathways. In order to evaluate pathway parameter advising, we compare its performance in avoiding implausible networks and reconstructing pathways from the NetPath database with other parameter selection methods across four pathway reconstruction algorithms. We also demonstrate how pathway parameter advising can guide reconstruction of an influenza host factor network. Pathway parameter advising is method agnostic; itmore »is applicable to any pathway reconstruction algorithm with tunable parameters.

    « less
  4. Abstract

    Assessments of the climate impacts of energy technologies and other emissions sources can depend strongly on the equivalency metric used to compare short- and long-lived greenhouse gas emissions. However, the consequences of metric design choices are not fully understood, and in practice, a single metric, the global warming potential (GWP), is used almost universally. Many metrics have been proposed and evaluated in recent decades, but questions still remain about which ones perform better and why. Here, we develop new insights on how the design of equivalency metrics can impact the outcomes of climate policies. We distill the equivalency metric problem into a few key design choices that determine the metric values and shapes seen across a wide range of different proposed metrics. We examine outcomes under a hypothetical 1.5 or 2C policy target and discuss extensions to other policies. Across policy contexts, the choice of time parameters is particularly important. Metrics that emphasize the immediate impacts of short-lived gases such as methane can reduce rates of climate change but may require more rapid technology changes. Differences in outcomes across metrics are more pronounced when fossil fuels, with or without carbon capture and storage, play a larger role in energymore »transitions. By identifying a small set of consequential design decisions, these insights can help make metric choices and energy transitions more deliberate and effective at mitigating climate change.

    « less
  5. Obeid, I. ; Selesnik, I. ; Picone, J. (Ed.)
    The Neuronix high-performance computing cluster allows us to conduct extensive machine learning experiments on big data [1]. This heterogeneous cluster uses innovative scheduling technology, Slurm [2], that manages a network of CPUs and graphics processing units (GPUs). The GPU farm consists of a variety of processors ranging from low-end consumer grade devices such as the Nvidia GTX 970 to higher-end devices such as the GeForce RTX 2080. These GPUs are essential to our research since they allow extremely compute-intensive deep learning tasks to be executed on massive data resources such as the TUH EEG Corpus [2]. We use TensorFlow [3] as the core machine learning library for our deep learning systems, and routinely employ multiple GPUs to accelerate the training process. Reproducible results are essential to machine learning research. Reproducibility in this context means the ability to replicate an existing experiment – performance metrics such as error rates should be identical and floating-point calculations should match closely. Three examples of ways we typically expect an experiment to be replicable are: (1) The same job run on the same processor should produce the same results each time it is run. (2) A job run on a CPU and GPU should producemore »identical results. (3) A job should produce comparable results if the data is presented in a different order. System optimization requires an ability to directly compare error rates for algorithms evaluated under comparable operating conditions. However, it is a difficult task to exactly reproduce the results for large, complex deep learning systems that often require more than a trillion calculations per experiment [5]. This is a fairly well-known issue and one we will explore in this poster. Researchers must be able to replicate results on a specific data set to establish the integrity of an implementation. They can then use that implementation as a baseline for comparison purposes. A lack of reproducibility makes it very difficult to debug algorithms and validate changes to the system. Equally important, since many results in deep learning research are dependent on the order in which the system is exposed to the data, the specific processors used, and even the order in which those processors are accessed, it becomes a challenging problem to compare two algorithms since each system must be individually optimized for a specific data set or processor. This is extremely time-consuming for algorithm research in which a single run often taxes a computing environment to its limits. Well-known techniques such as cross-validation [5,6] can be used to mitigate these effects, but this is also computationally expensive. These issues are further compounded by the fact that most deep learning algorithms are susceptible to the way computational noise propagates through the system. GPUs are particularly notorious for this because, in a clustered environment, it becomes more difficult to control which processors are used at various points in time. Another equally frustrating issue is that upgrades to the deep learning package, such as the transition from TensorFlow v1.9 to v1.13, can also result in large fluctuations in error rates when re-running the same experiment. Since TensorFlow is constantly updating functions to support GPU use, maintaining an historical archive of experimental results that can be used to calibrate algorithm research is quite a challenge. This makes it very difficult to optimize the system or select the best configurations. The overall impact of all of these issues described above is significant as error rates can fluctuate by as much as 25% due to these types of computational issues. Cross-validation is one technique used to mitigate this, but that is expensive since you need to do multiple runs over the data, which further taxes a computing infrastructure already running at max capacity. GPUs are preferred when training a large network since these systems train at least two orders of magnitude faster than CPUs [7]. Large-scale experiments are simply not feasible without using GPUs. However, there is a tradeoff to gain this performance. Since all our GPUs use the NVIDIA CUDA® Deep Neural Network library (cuDNN) [8], a GPU-accelerated library of primitives for deep neural networks, it adds an element of randomness into the experiment. When a GPU is used to train a network in TensorFlow, it automatically searches for a cuDNN implementation. NVIDIA’s cuDNN implementation provides algorithms that increase the performance and help the model train quicker, but they are non-deterministic algorithms [9,10]. Since our networks have many complex layers, there is no easy way to avoid this randomness. Instead of comparing each epoch, we compare the average performance of the experiment because it gives us a hint of how our model is performing per experiment, and if the changes we make are efficient. In this poster, we will discuss a variety of issues related to reproducibility and introduce ways we mitigate these effects. For example, TensorFlow uses a random number generator (RNG) which is not seeded by default. TensorFlow determines the initialization point and how certain functions execute using the RNG. The solution for this is seeding all the necessary components before training the model. This forces TensorFlow to use the same initialization point and sets how certain layers work (e.g., dropout layers). However, seeding all the RNGs will not guarantee a controlled experiment. Other variables can affect the outcome of the experiment such as training using GPUs, allowing multi-threading on CPUs, using certain layers, etc. To mitigate our problems with reproducibility, we first make sure that the data is processed in the same order during training. Therefore, we save the data from the last experiment and to make sure the newer experiment follows the same order. If we allow the data to be shuffled, it can affect the performance due to how the model was exposed to the data. We also specify the float data type to be 32-bit since Python defaults to 64-bit. We try to avoid using 64-bit precision because the numbers produced by a GPU can vary significantly depending on the GPU architecture [11-13]. Controlling precision somewhat reduces differences due to computational noise even though technically it increases the amount of computational noise. We are currently developing more advanced techniques for preserving the efficiency of our training process while also maintaining the ability to reproduce models. In our poster presentation we will demonstrate these issues using some novel visualization tools, present several examples of the extent to which these issues influence research results on electroencephalography (EEG) and digital pathology experiments and introduce new ways to manage such computational issues.« less