NullHop: A Flexible Convolutional Neural Network Accelerator Based on Sparse Representations of Feature Maps.

Convolutional neural networks (CNNs) have become the dominant neural network architecture for solving many state-of-the-art (SOA) visual processing tasks. Even though graphical processing units are most often used in training and deploying CNNs, their power efficiency is less than 10 GOp/s/W for single-frame runtime inference. We propose a flexible and efficient CNN accelerator architecture called NullHop that implements SOA CNNs useful for low-power and low-latency application scenarios. NullHop exploits the sparsity of neuron activations in CNNs to accelerate the computation and reduce memory requirements. The flexible architecture allows high utilization of available computing resources across kernel sizes ranging from 1×1 to 7×7 . NullHop can process up to 128 input and 128 output feature maps per layer in a single pass. We implemented the proposed architecture on a Xilinx Zynq field-programmable gate array (FPGA) platform and presented the results showing how our implementation reduces external memory transfers and compute time in five different CNNs ranging from small ones up to the widely known large VGG16 and VGG19 CNNs. Postsynthesis simulations using Mentor Modelsim in a 28-nm process with a clock frequency of 500 MHz show that the VGG19 network achieves over 450 GOp/s. By exploiting sparsity, NullHop achieves an efficiency of 368%, maintains over 98% utilization of the multiply-accumulate units, and achieves a power efficiency of over 3 TOp/s/W in a core area of 6.3 mm2. As further proof of NullHop's usability, we interfaced its FPGA implementation with a neuromorphic event camera for real-time interactive demonstrations.

Microinvasive squamous cell cervical carcinoma.

Several histologic tumor-related features are the key factors for further treatment planning in microinvasive cervical cancer (MIC) after conization. To better define the indications for conservative treatment of MIC we conducted a literature review for prognostic factors for MIC and we carried out a prospective observational study evaluating most important pathologic factors and the relationships between tumor and edges of the cone and incidence of recurrences. In our experience seven recurrences were observed. Two distinct groups of patients were identified with a clearance lower or higher of 10 and 8 mm for apical and lateral margin respectively. Depth of infiltration and even lymph-vascular involvement have been confirmed as the most important histologic parameters to be evaluated. Apical and lateral clearance of the tumor are significantly correlated with the recurrence rate. If an adequate lateral border of healthy tissue is present on the specimen, conization may be considered as definitive treatment of MIC.