Filtering by
- All Subjects: Neural Networks
- Creators: Vrudhula, Sarma
- Creators: Chen, Yinong
- Resource Type: Text
been a subject of great interest to policy and decision makers for the past 40 years.
Recent research indicates that while there exist specific shortages in specific disciplines
and areas of expertise in the private sector and the federal government, there is no
noticeable shortage in any STEM academic discipline, but rather a surplus of PhDs
vying for increasingly scarce tenure track positions. Despite the seeming availability
of industry and private sector jobs, recent PhDs still struggle to find employment in
those areas. I argue that the decades old narrative suggesting a shortage of STEM
PhDs in the US poses a threat to the value of the natural science PhD, and that
this narrative contributes significantly to why so many PhDs struggle to find career
employment in their fields. This study aims to address the following question: what is
the value of a STEM PhD outside academia? I begin with a critical review of existing
literature, and then analyze programmatic documents for STEM PhD programs at
ASU, interviews with industry employers, and an examination the public face of value
for these degrees. I then uncover the nature of the value alignment, value disconnect,
and value erosion in the ecosystem which produces and then employs STEM PhDs,
concluding with specific areas which merit special consideration in an effort to increase
the value of these degrees for all stakeholders involved.
Deep neural networks are composed of multiple layers that are compute/memory intensive. This makes it difficult to execute these algorithms real-time with low power consumption using existing general purpose computers. In this work, we propose hardware accelerators for a traditional AI algorithm based on random forest trees and two representative deep convolutional neural networks (AlexNet and VGG). With the proposed acceleration techniques, ~ 30x performance improvement was achieved compared to CPU for random forest trees. For deep CNNS, we demonstrate that much higher performance can be achieved with architecture space exploration using any optimization algorithms with system level performance and area models for hardware primitives as inputs and goal of minimizing latency with given resource constraints. With this method, ~30GOPs performance was achieved for Stratix V FPGA boards.
Hardware acceleration of DL algorithms alone is not always the most ecient way and sucient to achieve desired performance. There is a huge headroom available for performance improvement provided the algorithms are designed keeping in mind the hardware limitations and bottlenecks. This work achieves hardware-software co-optimization for Non-Maximal Suppression (NMS) algorithm. Using the proposed algorithmic changes and hardware architecture
With CMOS scaling coming to an end and increasing memory bandwidth bottlenecks, CMOS based system might not scale enough to accommodate requirements of more complicated and deeper neural networks in future. In this work, we explore RRAM crossbars and arrays as compact, high performing and energy efficient alternative to CMOS accelerators for deep learning training and inference. We propose and implement RRAM periphery read and write circuits and achieved ~3000x performance improvement in online dictionary learning compared to CPU.
This work also examines the realistic RRAM devices and their non-idealities. We do an in-depth study of the effects of RRAM non-idealities on inference accuracy when a pretrained model is mapped to RRAM based accelerators. To mitigate this issue, we propose Random Sparse Adaptation (RSA), a novel scheme aimed at tuning the model to take care of the faults of the RRAM array on which it is mapped. Our proposed method can achieve inference accuracy much higher than what traditional Read-Verify-Write (R-V-W) method could achieve. RSA can also recover lost inference accuracy 100x ~ 1000x faster compared to R-V-W. Using 32-bit high precision RSA cells, we achieved ~10% higher accuracy using fautly RRAM arrays compared to what can be achieved by mapping a deep network to an 32 level RRAM array with no variations.
This work presents an energy-efficient programmable application-specific integrated circuit (ASIC) accelerator for object detection. The proposed ASIC supports multi-class (face/traffic sign/car license plate/pedestrian), many-object (up to 50) in one image with different sizes (6 down-/11 up-scaling), and high accuracy (87% for face detection datasets). The proposed accelerator is composed of an integral channel detector with 2,000 classifiers for five rigid boosted templates to make a strong object detection. By jointly optimizing the algorithm and efficient hardware architecture, the prototype chip implemented in 65nm demonstrates real-time object detection of 20-50 frames/s with 22.5-181.7mW (0.54-1.75nJ/pixel) at 0.58-1.1V supply.
In this work, to reduce computation without accuracy degradation, an energy-efficient deep convolutional neural network (DCNN) accelerator is proposed based on a novel conditional computing scheme and integrates convolution with subsequent max-pooling operations. This way, the total number of bit-wise convolutions could be reduced by ~2x, without affecting the output feature values. This work also has been developing an optimized dataflow that exploits sparsity, maximizes data re-use and minimizes off-chip memory access, which can improve upon existing hardware works. The total off-chip memory access can be saved by 2.12x. Preliminary results of the proposed DCNN accelerator achieved a peak 7.35 TOPS/W for VGG-16 by post-layout simulation results in 40nm.
A number of recent efforts have attempted to design custom inference engine based on various approaches, including the systolic architecture, near memory processing, and in-meomry computing concept. This work evaluates a comprehensive comparison of these various approaches in a unified framework. This work also presents the proposed energy-efficient in-memory computing accelerator for deep neural networks (DNNs) by integrating many instances of in-memory computing macros with an ensemble of peripheral digital circuits, which supports configurable multibit activations and large-scale DNNs seamlessly while substantially improving the chip-level energy-efficiency. Proposed accelerator is fully designed in 65nm, demonstrating ultralow energy consumption for DNNs.
For my Honors Thesis, I decided to create an Artificial Intelligence Project to predict Fantasy NFL Football Points of players and team's defense. I created a Tensorflow Keras AI Regression model and created a Flask API that holds the AI model, and a Django Try-It Page for the user to use the model. These services are hosted on ASU's AWS service. In my Flask API, it actively gathers data from Pro-Football-Reference, then calculates the fantasy points. Let’s say the current year is 2022, then the model analyzes each player and trains on all data from available from 2000 to 2020 data, tests the data on 2021 data, and predicts for 2022 year. The Django Website asks the user to input the current year, then the user clicks the submit button runs the AI model, and the process explained earlier. Next, the user enters the player's name for the point prediction and the website predicts the last 5 rows with 4 being the previous fantasy points and the 5th row being the prediction.