More Like this

1. Neural Networks for Learning Counterfactual G-Invariances from Single Environments

   Mouli, S Chandra ; Ribeiro, Bruno ( April 2021 , ICLR)

   null (Ed.)

   Despite ---or maybe because of--- their astonishing capacity to fit data, neural networks are believed to have difficulties extrapolating beyond training data distribution. This work shows that, for extrapolations based on finite transformation groups, a model's inability to extrapolate is unrelated to its capacity. Rather, the shortcoming is inherited from a learning hypothesis: Examples not explicitly observed with infinitely many training examples have underspecified outcomes in the learner’s model. In order to endow neural networks with the ability to extrapolate over group transformations, we introduce a learning framework counterfactually-guided by the learning hypothesis that any group invariance to (known) transformation groups is mandatory even without evidence, unless the learner deems it inconsistent with the training data. 

   more » « less
2. Issues in the Reproducibility of Deep Learning Results

   https://doi.org/10.1109/SPMB47826.2019.9037840  

   Jean-Paul, S. ; Elseify, T. ; Obeid, I. ; Picone, J. ( December 2019 , IEEE Signal Processing in Medicine and Biology Symposium (SPMB))

   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 produce 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. 

   more » « less
3. Work in Progress: Interactive Introductory Online Modules on Wireless Communications and Radio-frequency Spectrum Sharing

   Dietrich, C. B. ; Polys, N. F. ; Reid, K. ; García Sheridan, J. A. ; Emenonye, D. ; Gomez, X. ; Tolley, J. ; Makin, C. ; Gaeddert, J. ( July 2021 , ASEE Annual Conference proceedings)

   1. Description of the objectives and motivation for the contribution to ECE education The demand for wireless data transmission capacity is increasing rapidly and this growth is expected to continue due to ongoing prevalence of cellular phones and new and emerging bandwidth-intensive applications that encompass high-definition video, unmanned aerial systems (UAS), intelligent transportation systems (ITS) including autonomous vehicles, and others. Meanwhile, vital military and public safety applications also depend on access to the radio frequency spectrum. To meet these demands, the US federal government is beginning to move from the proven but inefficient model of exclusive frequency assignments to a more-efficient, shared-spectrum approach in some bands of the radio frequency spectrum. A STEM workforce that understands the radio frequency spectrum and applications that use the spectrum is needed to further increase spectrum efficiency and cost-effectiveness of wireless systems over the next several decades to meet anticipated and unanticipated increases in wireless data capacity. 2. Relevant background including literature search examples if appropriate CISCO Systems’ annual survey indicates continued strong growth in demand for wireless data capacity. Meanwhile, undergraduate electrical and computer engineering courses in communication systems, electromagnetics, and networks tend to emphasize mathematical and theoretical fundamentals and higher-layer protocols, with less focus on fundamental concepts that are more specific to radio frequency wireless systems, including the physical and media access control layers of wireless communication systems and networks. An efficient way is needed to introduce basic RF system and spectrum concepts to undergraduate engineering students in courses such as those mentioned above who are unable to, or had not planned to take a full course in radio frequency / microwave engineering or wireless systems and networks. We have developed a series of interactive online modules that introduce concepts fundamental to wireless communications, the radio frequency spectrum, and spectrum sharing, and seek to present these concepts in context. The modules include interactive, JavaScript-based simulation exercises intended to reinforce the concepts that are presented in the modules through narrated slide presentations, text, and external links. Additional modules in development will introduce advanced undergraduate and graduate students and STEM professionals to configuration and programming of adaptive frequency-agile radios and spectrum management systems that can operate efficiently in congested radio frequency environments. Simulation exercises developed for the advanced modules allow both manual and automatic control of simulated radio links in timed, game-like simulations, and some exercises will enable students to select from among multiple pre-coded controller strategies and optionally edit the code before running the timed simulation. Additionally, we have developed infrastructure for running remote laboratory experiments that can also be embedded within the online modules, including a web-based user interface, an experiment management framework, and software defined radio (SDR) application software that runs in a wireless testbed initially developed for research. Although these experiments rely on limited hardware resources and introduce additional logistical considerations, they provide additional realism that may further challenge and motivate students. 3. Description of any assessment methods used to evaluate the effectiveness of the contribution, Each set of modules is preceded and followed by a survey. Each individual module is preceded by a quiz and followed by another quiz, with pre- and post-quiz questions drawn from the same pool. The pre-surveys allow students to opt in or out of having their survey and quiz results used anonymously in research. 4. Statement of results. The initial modules have been and are being used by three groups of students: (1) students in an undergraduate Introduction to Communication Systems course; (2) an interdisciplinary group of engineering students, including computer science students, who are participating in related undergraduate research project; and (3) students in a graduate-level communications course that includes both electrical and computer engineers. Analysis of results from the first group of students showed statistically significant increases from pre-quiz to post-quiz for each of four modules on fundamental wireless communication concepts. Results for the other students have not yet been analyzed, but also appear to show substantial pre-quiz to post-quiz increases in mean scores. 

   more » « less
4. Adar: Adversarial Activity Recognition in Wearables

   https://doi.org/10.1109/ICCAD45719.2019.8942124  

   Sah, Ramesh Kumar ; Ghasemzadeh, Hassan ( November 2019 , Proceedings of the International Conference on Computer-Aided Design (ICCAD 2019))

   Recent advances in machine learning and deep neural networks have led to the realization of many important applications in the area of personalized medicine. Whether it is detecting activities of daily living or analyzing images for cancerous cells, machine learning algorithms have become the dominant choice for such emerging applications. In particular, the state-of-the-art algorithms used for human activity recognition (HAR) using wearable inertial sensors utilize machine learning algorithms to detect health events and to make predictions from sensor data. Currently, however, there remains a gap in research on whether or not and how activity recognition algorithms may become the subject of adversarial attacks. In this paper, we take the first strides on (1) investigating methods of generating adversarial example in the context of HAR systems; (2) studying the vulnerability of activity recognition models to adversarial examples in feature and signal domain; and (3) investigating the effects of adversarial training on HAR systems. We introduce Adar, a novel computational framework for optimization-driven creation of adversarial examples in sensor-based activity recognition systems. Through extensive analysis based on real sensor data collected with human subjects, we found that simple evasion attacks are able to decrease the accuracy of a deep neural network from 95.1% to 3.4% and from 93.1% to 16.8% in the case of a convolutional neural network. With adversarial training, the robustness of the deep neural network increased on the adversarial examples by 49.1% in the worst case while the accuracy on clean samples decreased by 13.2%. 

   more » « less
5. How neural networks extrapolate: from feedforward to graph neural networks

   Xu, Keyulu ; Zhang, Mozhi ; Li, Jingling ; Du, Simon S. ; Kawarabayashi, Ken-ichi ; Jegelka, Stefanie ( January 2021 , International Conference on Learning Representations (ICLR))

   null (Ed.)

   We study how neural networks trained by gradient descent extrapolate, i.e., what they learn outside the support of the training distribution. Previous works report mixed empirical results when extrapolating with neural networks: while feedforward neural networks, a.k.a. multilayer perceptrons (MLPs), do not extrapolate well in certain simple tasks, Graph Neural Networks (GNNs) – structured networks with MLP modules – have shown some success in more complex tasks. Working towards a theoretical explanation, we identify conditions under which MLPs and GNNs extrapolate well. First, we quantify the observation that ReLU MLPs quickly converge to linear functions along any direction from the origin, which implies that ReLU MLPs do not extrapolate most nonlinear functions. But, they can provably learn a linear target function when the training distribution is sufficiently “diverse”. Second, in connection to analyzing the successes and limitations of GNNs, these results suggest a hypothesis for which we provide theoretical and empirical evidence: the success of GNNs in extrapolating algorithmic tasks to new data (e.g., larger graphs or edge weights) relies on encoding task-specific non-linearities in the architecture or features. Our theoretical analysis builds on a connection of over-parameterized networks to the neural tangent kernel. Empirically, our theory holds across different training settings. 

   more » « less