Spiking neuralnetworks (SNNs) have recently been used as a computational model for applications such as deep learning, image recognition and machinelearning. Similar to the biological brain, SNN neurons depend on th...
详细信息
ISBN:
(纸本)9798350372977;9798350372984
Spiking neuralnetworks (SNNs) have recently been used as a computational model for applications such as deep learning, image recognition and machinelearning. Similar to the biological brain, SNN neurons depend on the membrane level to fire an output. If the level exceeds a specified threshold, the neuron sends an output to activate next neurons. This leads to an unbalanced workload among the neurons. The dynamically-changing membrane level is stored inside a neuron. In hardware, this storage can be implemented as a register or on-chip memory, which determines the amount of consumed resources and, in turns, affects the network scalability. SNN accelerators have recently been implemented on UltraScale FPGAs devices for high-performance purposes. On-chip memories on these devices are classified as distributed memory, Block RAMs (BRAMs) and Ultra RAMs (URAMs). In this paper, we explored the impact of using different on-chip memories to store the membrane level of SNN neurons. We implemented a parameterizable SpIking neural networK (SINK) accelerator where the network capacity and weight width are parameters. SINK has the ability to run in four different modes based on the memory type. We ran SINK on UltraScale Zync104 FPGA device and measure the utilization of the hardware resources (LUTs), registers, memory, power consumption and performance. The results show that URAM can be the best fit to store the membrane level, since it used 30%, 11% and 2% less LUTs, Regs, and power 2% respectively comparing with BRAM and distributed memory
Pneumonia, an infectious lung condition caused by bacteria, viruses, or other microorganisms, significantly impacts both pediatric and geriatric populations. The COVID-19 pandemic has underscored the necessity for swi...
详细信息
In the billions of faces that are shaped by thousands of different cultures and ethnicities, one thing remains universal: the way emotions are expressed. To take the next step in human-machine interactions, a machine ...
详细信息
ISBN:
(数字)9781510662117
ISBN:
(纸本)9781510662100;9781510662117
In the billions of faces that are shaped by thousands of different cultures and ethnicities, one thing remains universal: the way emotions are expressed. To take the next step in human-machine interactions, a machine must be able to clarify facial emotions. Allowing machines to recognize micro-expressions gives them a deeper dive into a person's true feelings at an instant which allows designers to create more empathetic machines that will take human emotion into account while making optimal decisions;e.g., these machines will be potentially able to detect dangerous situations, alert caregivers to challenges, and provide appropriate responses. Micro-expressions are involuntary and transient facial expressions capable of revealing genuine emotions. We propose to design and train a set of neural network (NN) models capable of micro-expression recognition in real-time applications. Different NN models are explored and compared in this study to design a hybrid deep learning model by combining a convolutional neural network (CNN), a recurrent neural network (RNN, e.g., long short-term memory [LSTM]), and a vision transformer. The CNN can extract spatial features (of a neighborhood within an image) whereas the LSTM can summarize temporal features. In addition, a transformer with an attention mechanism can capture sparse spatial relations residing an image or between frames in a video clip. The inputs of the model are short facial videos, while the outputs are the micro-expressions gleaned from the videos. The deep learning models are trained and tested with publicly available facial micro-expression datasets to recognize different micro-expressions (e.g., happiness, fear, anger, surprise, disgust, sadness). The results of our proposed models are compared with that of literature-reported methods tested on the same datasets. The proposed hybrid models perform the best.
In this paper, we examine and propose a super resolution technique for Magnetic Particle Imaging (MPI) developed using transfer learning and sparse transforms. MPI is a new modality for medical imaging that relies on ...
详细信息
ISBN:
(纸本)9798350387186;9798350387179
In this paper, we examine and propose a super resolution technique for Magnetic Particle Imaging (MPI) developed using transfer learning and sparse transforms. MPI is a new modality for medical imaging that relies on tracking of super-paramagnetic nanoparticles and offers a superior alternative to medical imaging methods such as MRI and X-Ray in some applications. Like other imaging modalities, and despite being faster than other methods, MPI has limitations in spatial and temporal resolution. Super resolution techniques can considerably widen the applications of MPI, for real-time and in-vivo scanning and analysis. The current lack of MPI data prevents direct development of machinelearning and deep neural network (DNN) based methods for super resolution. To overcome this issue, we study the application of transfer learning and show that utilizing some existing large datasets of natural images, and then retraining on a small dataset of MPI images allows for successful development of a super resolution DNN. We also propose to use sparsifying transforms in the early stages of the DNN, for improved quality of the inferred increased resolution image. We show that for such networks a 4-5 dB improvement is achieved over the common bicubic interpolation techniques for super resolution.
Interpreting the results of Convolutional neuralnetworks remains a challenging task. Quantitative evaluations apart from precision, recall, and their extensions are rare and usually do not cover the necessary aspects...
详细信息
ISBN:
(纸本)9781728198354
Interpreting the results of Convolutional neuralnetworks remains a challenging task. Quantitative evaluations apart from precision, recall, and their extensions are rare and usually do not cover the necessary aspects of specific applications. In this work, a methodology based on the intrinsic dimensionality of the image space and latent space in multiple layers is presented. This methodology has been used in other literature for classification but is leveraged to object detection where the interpretation of the results is more complex. The suitability of the intrinsic dimensionality is evaluated first for general augmentation techniques in multiple datasets and with multiple networks and later on a specific use case with multiple disturbances included. With the help of the intrinsic dimensionality, conclusions about the robustness can be drawn which are not apparent from the precision and the suitability of the methodology as an auxiliary quantifiable metric therefore shown.
The type and volume of malware is increasing rapidly. In this work we present an improved approach for malware classification from the android applications (or apps) using image features. The android apps are converte...
详细信息
In the era of digital imagery, there is a great interest in finding new and creative ways to express ourselves and make our images look beautiful. One such fascinating method is cartoonization, a process that transfor...
详细信息
In recent years, we have observed the growing vulnerability of deep neuralnetworks (DNNs) to adversarial attacks, challenging the forefronts of machinelearning. Adver-sarial machinelearning has emerged as a crucial...
详细信息
Homologous anatomical landmarks between medical scans are instrumental in quantitative assessment of image registration quality in various clinical applications, such as MRI-ultrasound registration for tissue shift co...
详细信息
ISBN:
(纸本)9783031439957;9783031439964
Homologous anatomical landmarks between medical scans are instrumental in quantitative assessment of image registration quality in various clinical applications, such as MRI-ultrasound registration for tissue shift correction in ultrasound-guided brain tumor resection. While manually identified landmark pairs between MRI and ultrasound (US) have greatly facilitated the validation of different registration algorithms for the task, the procedure requires significant expertise, labor, and time, and can be prone to inter- and intra-rater inconsistency. So far, many traditional and machinelearning approaches have been presented for anatomical landmark detection, but they primarily focus on mono-modal applications. Unfortunately, despite the clinical needs, inter-modal/contrast landmark detection has very rarely been attempted. Therefore, we propose a novel contrastive learning framework to detect corresponding landmarks between MRI and intra-operative US scans in neurosurgery. Specifically, two convolutional neuralnetworks were trained jointly to encode image features in MRI and US scans to help match the US image patch that contain the corresponding landmarks in the MRI. We developed and validated the technique using the public RESECT database. With a mean landmark detection accuracy of 5.88 +/- 4.79 mm against 18.78 +/- 4.77 mm with SIFT features, the proposed method offers promising results for MRI-US landmark detection in neurosurgical applications for the first time.
Deep learning, a cornerstone of artificial intelligence (AI), has revolutionized a number of fields, including self-driving cars, image recognition, and intelligent medical applications. These developments have aided ...
详细信息
暂无评论