arXiv subjects

Priyadarshini Panda

Publications and source records attributed to Priyadarshini Panda.

At least 109 records · Page 6Linked to original sources

arXiv2020

Relevant-features based Auxiliary Cells for Energy Efficient Detection of Natural Errors

Deep neural networks have demonstrated state-of-the-art performance on many classification tasks. However, they have no inherent capability to recognize when their predictions are wrong. There have been several efforts in the recent past to detect natural errors but the suggested mechanisms pose additional energy requirements. To address this issue, we propose an ensemble of classifiers at hidden layers to enable energy efficient detection of natural errors. In particular, we append Relevant-features based Auxiliary Cells (RACs) which are class specific binary linear classifiers trained on relevant features. The consensus of RACs is used to detect natural errors. Based on combined confidence of RACs, classification can be terminated early, thereby resulting in energy efficient detection. We demonstrate the effectiveness of our technique on various image classification datasets such as CIFAR-10, CIFAR-100 and Tiny-ImageNet.

cs.LG↗

arXiv2020

A Low Effort Approach to Structured CNN Design Using PCA

Deep learning models hold state of the art performance in many fields, yet their design is still based on heuristics or grid search methods that often result in overparametrized networks. This work proposes a method to analyze a trained network and deduce an optimized, compressed architecture that preserves accuracy while keeping computational costs tractable. Model compression is an active field of research that targets the problem of realizing deep learning models in hardware. However, most pruning methodologies tend to be experimental, requiring large compute and time intensive iterations of retraining the entire network. We introduce structure into model design by proposing a single shot analysis of a trained network that serves as a first order, low effort approach to dimensionality reduction, by using PCA (Principal Component Analysis). The proposed method simultaneously analyzes the activations of each layer and considers the dimensionality of the space described by the filters generating these activations. It optimizes the architecture in terms of number of layers, and number of filters per layer without any iterative retraining procedures, making it a viable, low effort technique to design efficient networks. We demonstrate the proposed methodology on AlexNet and VGG style networks on the CIFAR-10, CIFAR-100 and ImageNet datasets, and successfully achieve an optimized architecture with a reduction of up to 3.8X and 9X in the number of operations and parameters respectively, while trading off less than 1% accuracy. We also apply the method to MobileNet, and achieve 1.7X and 3.9X reduction in the number of operations and parameters respectively, while improving accuracy by almost one percentage point.

cs.CV↗

arXiv2019

Towards Scalable, Efficient and Accurate Deep Spiking Neural Networks with Backward Residual Connections, Stochastic Softmax and Hybridization

Spiking Neural Networks (SNNs) may offer an energy-efficient alternative for implementing deep learning applications. In recent years, there have been several proposals focused on supervised (conversion, spike-based gradient descent) and unsupervised (spike timing dependent plasticity) training methods to improve the accuracy of SNNs on large-scale tasks. However, each of these methods suffer from scalability, latency and accuracy limitations. In this paper, we propose novel algorithmic techniques of modifying the SNN configuration with backward residual connections, stochastic softmax and hybrid artificial-and-spiking neuronal activations to improve the learning ability of the training methodologies to yield competitive accuracy, while, yielding large efficiency gains over their artificial counterparts. Note, artificial counterparts refer to conventional deep learning/artificial neural networks. Our techniques apply to VGG/Residual architectures, and are compatible with all forms of training methodologies. Our analysis reveals that the proposed solutions yield near state-of-the-art accuracy with significant energy-efficiency and reduced parameter overhead translating to hardware improvements on complex visual recognition tasks, such as, CIFAR10, Imagenet datatsets.

cs.CV↗

arXiv2019

Tree-CNN: A Hierarchical Deep Convolutional Neural Network for Incremental Learning

Over the past decade, Deep Convolutional Neural Networks (DCNNs) have shown remarkable performance in most computer vision tasks. These tasks traditionally use a fixed dataset, and the model, once trained, is deployed as is. Adding new information to such a model presents a challenge due to complex training issues, such as "catastrophic forgetting", and sensitivity to hyper-parameter tuning. However, in this modern world, data is constantly evolving, and our deep learning models are required to adapt to these changes. In this paper, we propose an adaptive hierarchical network structure composed of DCNNs that can grow and learn as new data becomes available. The network grows in a tree-like fashion to accommodate new classes of data, while preserving the ability to distinguish the previously trained classes. The network organizes the incrementally available data into feature-driven super-classes and improves upon existing hierarchical CNN models by adding the capability of self-growth. The proposed hierarchical model, when compared against fine-tuning a deep network, achieves significant reduction of training effort, while maintaining competitive accuracy on CIFAR-10 and CIFAR-100.

cs.CV↗

arXiv2019

Implicit Generative Modeling of Random Noise during Training for Adversarial Robustness

We introduce a Noise-based prior Learning (NoL) approach for training neural networks that are intrinsically robust to adversarial attacks. We find that the implicit generative modeling of random noise with the same loss function used during posterior maximization, improves a model's understanding of the data manifold furthering adversarial robustness. We evaluate our approach's efficacy and provide a simplistic visualization tool for understanding adversarial data, using Principal Component Analysis. Our analysis reveals that adversarial robustness, in general, manifests in models with higher variance along the high-ranked principal components. We show that models learnt with our approach perform remarkably well against a wide-range of attacks. Furthermore, combining NoL with state-of-the-art adversarial training extends the robustness of a model, even beyond what it is adversarially trained for, in both white-box and black-box attack scenarios.

cs.LG↗

arXiv2019

Synthesizing Images from Spatio-Temporal Representations using Spike-based Backpropagation

Spiking neural networks (SNNs) offer a promising alternative to current artificial neural networks to enable low-power event-driven neuromorphic hardware. Spike-based neuromorphic applications require processing and extracting meaningful information from spatio-temporal data, represented as series of spike trains over time. In this paper, we propose a method to synthesize images from multiple modalities in a spike-based environment. We use spiking auto-encoders to convert image and audio inputs into compact spatio-temporal representations that is then decoded for image synthesis. For this, we use a direct training algorithm that computes loss on the membrane potential of the output layer and back-propagates it by using a sigmoid approximation of the neuron's activation function to enable differentiability. The spiking autoencoders are benchmarked on MNIST and Fashion-MNIST and achieve very low reconstruction loss, comparable to ANNs. Then, spiking autoencoders are trained to learn meaningful spatio-temporal representations of the data, across the two modalities - audio and visual. We synthesize images from audio in a spike-based environment by first generating, and then utilizing such shared multi-modal spatio-temporal representations. Our audio to image synthesis model is tested on the task of converting TI-46 digits audio samples to MNIST images. We are able to synthesize images with high fidelity and the model achieves competitive performance against ANNs.

cs.NE↗

arXiv2019

Evaluating the Stability of Recurrent Neural Models during Training with Eigenvalue Spectra Analysis

We analyze the stability of recurrent networks, specifically, reservoir computing models during training by evaluating the eigenvalue spectra of the reservoir dynamics. To circumvent the instability arising in examining a closed loop reservoir system with feedback, we propose to break the closed loop system. Essentially, we unroll the reservoir dynamics over time while incorporating the feedback effects that preserve the overall temporal integrity of the system. We evaluate our methodology for fixed point and time varying targets with least squares regression and FORCE training, respectively. Our analysis establishes eigenvalue spectra (which is, shrinking of spectral circle as training progresses) as a valid and effective metric to gauge the convergence of training as well as the convergence of the chaotic activity of the reservoir toward stable states.

cs.NE↗

arXiv2019

A Comprehensive Analysis on Adversarial Robustness of Spiking Neural Networks

In this era of machine learning models, their functionality is being threatened by adversarial attacks. In the face of this struggle for making artificial neural networks robust, finding a model, resilient to these attacks, is very important. In this work, we present, for the first time, a comprehensive analysis of the behavior of more bio-plausible networks, namely Spiking Neural Network (SNN) under state-of-the-art adversarial tests. We perform a comparative study of the accuracy degradation between conventional VGG-9 Artificial Neural Network (ANN) and equivalent spiking network with CIFAR-10 dataset in both whitebox and blackbox setting for different types of single-step and multi-step FGSM (Fast Gradient Sign Method) attacks. We demonstrate that SNNs tend to show more resiliency compared to ANN under black-box attack scenario. Additionally, we find that SNN robustness is largely dependent on the corresponding training mechanism. We observe that SNNs trained by spike-based backpropagation are more adversarially robust than the ones obtained by ANN-to-SNN conversion rules in several whitebox and blackbox scenarios. Finally, we also propose a simple, yet, effective framework for crafting adversarial attacks from SNNs. Our results suggest that attacks crafted from SNNs following our proposed method are much stronger than those crafted from ANNs.

cs.NE↗

arXiv2019

Discretization based Solutions for Secure Machine Learning against Adversarial Attacks

Adversarial examples are perturbed inputs that are designed (from a deep learning network's (DLN) parameter gradients) to mislead the DLN during test time. Intuitively, constraining the dimensionality of inputs or parameters of a network reduces the 'space' in which adversarial examples exist. Guided by this intuition, we demonstrate that discretization greatly improves the robustness of DLNs against adversarial attacks. Specifically, discretizing the input space (or allowed pixel levels from 256 values or 8-bit to 4 values or 2-bit) extensively improves the adversarial robustness of DLNs for a substantial range of perturbations for minimal loss in test accuracy. Furthermore, we find that Binary Neural Networks (BNNs) and related variants are intrinsically more robust than their full precision counterparts in adversarial scenarios. Combining input discretization with BNNs furthers the robustness even waiving the need for adversarial training for certain magnitude of perturbation values. We evaluate the effect of discretization on MNIST, CIFAR10, CIFAR100 and Imagenet datasets. Across all datasets, we observe maximal adversarial resistance with 2-bit input discretization that incurs an adversarial accuracy loss of just ~1-2% as compared to clean test accuracy.

cs.LG↗

arXiv2018

Exploiting Inherent Error-Resiliency of Neuromorphic Computing to achieve Extreme Energy-Efficiency through Mixed-Signal Neurons

Neuromorphic computing, inspired by the brain, promises extreme efficiency for certain classes of learning tasks, such as classification and pattern recognition. The performance and power consumption of neuromorphic computing depends heavily on the choice of the neuron architecture. Digital neurons (Dig-N) are conventionally known to be accurate and efficient at high speed, while suffering from high leakage currents from a large number of transistors in a large design. On the other hand, analog/mixed-signal neurons are prone to noise, variability and mismatch, but can lead to extremely low-power designs. In this work, we will analyze, compare and contrast existing neuron architectures with a proposed mixed-signal neuron (MS-N) in terms of performance, power and noise, thereby demonstrating the applicability of the proposed mixed-signal neuron for achieving extreme energy-efficiency in neuromorphic computing. The proposed MS-N is implemented in 65 nm CMOS technology and exhibits > 100X better energy-efficiency across all frequencies over two traditional digital neurons synthesized in the same technology node. We also demonstrate that the inherent error-resiliency of a fully connected or even convolutional neural network (CNN) can handle the noise as well as the manufacturing non-idealities of the MS-N up to certain degrees. Notably, a system-level implementation on MNIST datasets exhibits a worst-case increase in classification error by 2.1% when the integrated noise power in the bandwidth is ~ 0.1 uV2, along with +-3σ amount of variation and mismatch introduced in the transistor parameters for the proposed neuron with 8-bit precision.

cs.ET↗

arXiv2018

ASP: Learning to Forget with Adaptive Synaptic Plasticity in Spiking Neural Networks

A fundamental feature of learning in animals is the "ability to forget" that allows an organism to perceive, model and make decisions from disparate streams of information and adapt to changing environments. Against this backdrop, we present a novel unsupervised learning mechanism ASP (Adaptive Synaptic Plasticity) for improved recognition with Spiking Neural Networks (SNNs) for real time on-line learning in a dynamic environment. We incorporate an adaptive weight decay mechanism with the traditional Spike Timing Dependent Plasticity (STDP) learning to model adaptivity in SNNs. The leak rate of the synaptic weights is modulated based on the temporal correlation between the spiking patterns of the pre- and post-synaptic neurons. This mechanism helps in gradual forgetting of insignificant data while retaining significant, yet old, information. ASP, thus, maintains a balance between forgetting and immediate learning to construct a stable-plastic self-adaptive SNN for continuously changing inputs. We demonstrate that the proposed learning methodology addresses catastrophic forgetting while yielding significantly improved accuracy over the conventional STDP learning method for digit recognition applications. Additionally, we observe that the proposed learning model automatically encodes selective attention towards relevant features in the input data while eliminating the influence of background noise (or denoising) further improving the robustness of the ASP learning.

cs.NE↗

arXiv2018

Chaos-guided Input Structuring for Improved Learning in Recurrent Neural Networks

Anatomical studies demonstrate that brain reformats input information to generate reliable responses for performing computations. However, it remains unclear how neural circuits encode complex spatio-temporal patterns. We show that neural dynamics are strongly influenced by the phase alignment between the input and the spontaneous chaotic activity. Input structuring along the dominant chaotic projections causes the chaotic trajectories to become stable channels (or attractors), hence, improving the computational capability of a recurrent network. Using mean field analysis, we derive the impact of input structuring on the overall stability of attractors formed. Our results indicate that input alignment determines the extent of intrinsic noise suppression and hence, alters the attractor state stability, thereby controlling the network's inference ability.

q-bio.NC↗

arXiv2017

Magnetic Tunnel Junction Mimics Stochastic Cortical Spiking Neurons

Brain-inspired computing architectures attempt to mimic the computations performed in the neurons and the synapses in the human brain in order to achieve its efficiency in learning and cognitive tasks. In this work, we demonstrate the mapping of the probabilistic spiking nature of pyramidal neurons in the cortex to the stochastic switching behavior of a Magnetic Tunnel Junction in presence of thermal noise. We present results to illustrate the efficiency of neuromorphic systems based on such probabilistic neurons for pattern recognition tasks in presence of lateral inhibition and homeostasis. Such stochastic MTJ neurons can also potentially provide a direct mapping to the probabilistic computing elements in Belief Networks for performing regenerative tasks.

cs.ET↗

arXiv2017

Voltage-Driven Domain-Wall Motion based Neuro-Synaptic Devices for Dynamic On-line Learning

Conventional von-Neumann computing models have achieved remarkable feats for the past few decades. However, they fail to deliver the required efficiency for certain basic tasks like image and speech recognition when compared to biological systems. As such, taking cues from biological systems, novel computing paradigms are being explored for efficient hardware implementations of recognition/classification tasks. The basic building blocks of such neuromorphic systems are neurons and synapses. Towards that end, we propose a leaky-integrate-fire (LIF) neuron and a programmable non-volatile synapse using domain wall motion induced by magneto-electric effect. Due to a strong elastic pinning between the ferro-magnetic domain wall (FM-DW) and the underlying ferro-electric domain wall (FE-DW), the FM-DW gets dragged by the FE-DW on application of a voltage pulse. The fact that FE materials are insulators allows for pure voltage-driven FM-DW motion, which in turn can be used to mimic the behaviors of biological spiking neurons and synapses. The voltage driven nature of the proposed devices allows energy-efficient operation. A detailed device to system level simulation framework based on micromagnetic simulations has been developed to analyze the feasibility of the proposed neuro-synaptic devices. We also demonstrate that the energy-efficient voltage-controlled behavior of the proposed devices make them suitable for dynamic on-line and lifelong learning in spiking neural networks (SNNs).

cs.ET↗

arXiv2017

An Energy-Efficient Mixed-Signal Neuron for Inherently Error-Resilient Neuromorphic Systems

This work presents the design and analysis of a mixed-signal neuron (MS-N) for convolutional neural networks (CNN) and compares its performance with a digital neuron (Dig-N) in terms of operating frequency, power and noise. The circuit-level implementation of the MS-N in 65 nm CMOS technology exhibits 2-3 orders of magnitude better energy-efficiency over Dig-N for neuromorphic computing applications - especially at low frequencies due to the high leakage currents from many transistors in Dig-N. The inherent error-resiliency of CNN is exploited to handle the thermal and flicker noise of MS-N. A system-level analysis using a cohesive circuit-algorithmic framework on MNIST and CIFAR-10 datasets demonstrate an increase of 3% in worst-case classification error for MNIST when the integrated noise power in the bandwidth is ~ 1 μV2.

cs.ET↗

arXiv2017

Learning to Recognize Actions from Limited Training Examples Using a Recurrent Spiking Neural Model

A fundamental challenge in machine learning today is to build a model that can learn from few examples. Here, we describe a reservoir based spiking neural model for learning to recognize actions with a limited number of labeled videos. First, we propose a novel encoding, inspired by how microsaccades influence visual perception, to extract spike information from raw video data while preserving the temporal correlation across different frames. Using this encoding, we show that the reservoir generalizes its rich dynamical activity toward signature action/movements enabling it to learn from few training examples. We evaluate our approach on the UCF-101 dataset. Our experiments demonstrate that our proposed reservoir achieves 81.3%/87% Top-1/Top-5 accuracy, respectively, on the 101-class data while requiring just 8 video examples per class for training. Our results establish a new benchmark for action recognition from limited video examples for spiking neural models while yielding competetive accuracy with respect to state-of-the-art non-spiking neural models.

cs.NE↗

arXiv2017

STDP Based Pruning of Connections and Weight Quantization in Spiking Neural Networks for Energy Efficient Recognition

Spiking Neural Networks (SNNs) with a large number of weights and varied weight distribution can be difficult to implement in emerging in-memory computing hardware due to the limitations on crossbar size (implementing dot product), the constrained number of conductance levels in non-CMOS devices and the power budget. We present a sparse SNN topology where non-critical connections are pruned to reduce the network size and the remaining critical synapses are weight quantized to accommodate for limited conductance levels. Pruning is based on the power law weight-dependent Spike Timing Dependent Plasticity (STDP) model; synapses between pre- and post-neuron with high spike correlation are retained, whereas synapses with low correlation or uncorrelated spiking activity are pruned. The weights of the retained connections are quantized to the available number of conductance levels. The process of pruning non-critical connections and quantizing the weights of critical synapses is performed at regular intervals during training. We evaluated our sparse and quantized network on MNIST dataset and on a subset of images from Caltech-101 dataset. The compressed topology achieved a classification accuracy of 90.1% (91.6%) on the MNIST (Caltech-101) dataset with 3.1x (2.2x) and 4x (2.6x) improvement in energy and area, respectively. The compressed topology is energy and area efficient while maintaining the same classification accuracy of a 2-layer fully connected SNN topology.

cs.NE↗

arXiv2017

Gabor Filter Assisted Energy Efficient Fast Learning Convolutional Neural Networks

Convolutional Neural Networks (CNN) are being increasingly used in computer vision for a wide range of classification and recognition problems. However, training these large networks demands high computational time and energy requirements; hence, their energy-efficient implementation is of great interest. In this work, we reduce the training complexity of CNNs by replacing certain weight kernels of a CNN with Gabor filters. The convolutional layers use the Gabor filters as fixed weight kernels, which extracts intrinsic features, with regular trainable weight kernels. This combination creates a balanced system that gives better training performance in terms of energy and time, compared to the standalone CNN (without any Gabor kernels), in exchange for tolerable accuracy degradation. We show that the accuracy degradation can be mitigated by partially training the Gabor kernels, for a small fraction of the total training cycles. We evaluated the proposed approach on 4 benchmark applications. Simple tasks like face detection and character recognition (MNIST and TiCH), were implemented using LeNet architecture. While a more complex task of object recognition (CIFAR10) was implemented on a state of the art deep CNN (Network in Network) architecture. The proposed approach yields 1.31-1.53x improvement in training energy in comparison to conventional CNN implementation. We also obtain improvement up to 1.4x in training time, up to 2.23x in storage requirements, and up to 2.2x in memory access energy. The accuracy degradation suffered by the approximate implementations is within 0-3% of the baseline.

cs.NE↗