適用於固態... 硬碟之強健邏輯實體位址映射錯誤偵測方案 (A Robust Error Detection Scheme for Logical-to-Physical Address Mapping in Solid-State Drives)more

適用於固態硬碟之強健邏輯實體位址映射錯誤偵測方案 (A Robust Error Detection Scheme for Logical-to-Physical Address Mapping in Solid-State Drives)

用於晶片內... 網路AI資料傳輸的區塊式自適應表格近似通訊機制 (Block-wise Adaptive Table-based Approximate Communication for AI Data Transmission in Network-on-Chip)more

用於晶片內網路AI資料傳輸的區塊式自適應表格近似通訊機制 (Block-wise Adaptive Table-based Approximate Communication for AI Data Transmission in Network-on-Chip)

利用對角線... 多播器縱橫向多播與 Bank 鄰近核心存取分配，加速 Tenstorrent Tensix 架構上的 DRAM 與核心間的存取 (Accelerating DRAM-to-Core Access Using Diagonal Multicaster for Row and Column Multicasting and Bank-local Core Access Allocation on Tenstorrent’s Tensix-based Architecture)more

利用對角線多播器縱橫向多播與 Bank 鄰近核心存取分配，加速 Tenstorrent Tensix 架構上的 DRAM 與核心間的存取 (Accelerating DRAM-to-Core Access Using Diagonal Multicaster for Row and Column Multicasting and Bank-local Core Access Allocation on Tenstorrent’s Tensix-based Architecture)

時間差分為... 基礎的像素層級 N:M 與冷通道剪枝應用於擴散模型圖像生成 (Diffusion Image Generation with Temporal-Difference-Based Pixel-Wise N:M and Cold-Channel Pruning)more

時間差分為基礎的像素層級 N:M 與冷通道剪枝應用於擴散模型圖像生成 (Diffusion Image Generation with Temporal-Difference-Based Pixel-Wise N:M and Cold-Channel Pruning)

利用引導差... 分計算與位元切片架構加速無分類器引導的圖像生成 (Accelerating Classifier-Free Guided Image Generation using Guidance Differential Computing on a Bit-Slice Architecture)more

利用引導差分計算與位元切片架構加速無分類器引導的圖像生成 (Accelerating Classifier-Free Guided Image Generation using Guidance Differential Computing on a Bit-Slice Architecture)

基於 MX... INT 之指數與正負號近似的自注意力剪枝：應用於視覺與擴散轉換器 (MXINT-Based Exponent-Sign Approximation for Self-Attention Pruning in Vision and Diffusion Transformers)more

基於 MXINT 之指數與正負號近似的自注意力剪枝：應用於視覺與擴散轉換器 (MXINT-Based Exponent-Sign Approximation for Self-Attention Pruning in Vision and Diffusion Transformers)

基於ARC... 嵌入式平台之無人機飛行控制系統軟硬體協同設計與功能驗證 (Hardware-Software Co-Design and Functional Verification of UAV Flight Control System Based on ARC Embedded Platform)more

基於ARC嵌入式平台之無人機飛行控制系統軟硬體協同設計與功能驗證 (Hardware-Software Co-Design and Functional Verification of UAV Flight Control System Based on ARC Embedded Platform)

基於非線性... 模型預測控制的深度強化學習之自適應權重調整運用在液體儲槽系統中 (Adaptive Tuning of Action Weights in Deep Reinforcement Learning Models Based on Nonlinear Model Predictive Control for Liquid Tank Systems)more

基於非線性模型預測控制的深度強化學習之自適應權重調整運用在液體儲槽系統中 (Adaptive Tuning of Action Weights in Deep Reinforcement Learning Models Based on Nonlinear Model Predictive Control for Liquid Tank Systems)

逐層Top... -K圖案和核剪枝在物件偵測模型上的應用 (Layer-by-layer top-k pattern and kernel-wise pruning on object detection model)more

逐層Top-K圖案和核剪枝在物件偵測模型上的應用 (Layer-by-layer top-k pattern and kernel-wise pruning on object detection model)

應用於視覺... 轉換器的頻道方向性和粗質指數偏差的自適應浮點數量化 (Adaptive Floating Point Quantization with Coarse-grained and Channel-wise Exponent Bias for Vision Transformer)more

應用於視覺轉換器的頻道方向性和粗質指數偏差的自適應浮點數量化 (Adaptive Floating Point Quantization with Coarse-grained and Channel-wise Exponent Bias for Vision Transformer)

使用離線錯... 誤分析恢復記憶體內計算裁剪機制的準確度損失 (Recovering Accuracy Loss of Compute-in-Memory Clipping Mechanism using Offline Error Analyzing)more

使用離線錯誤分析恢復記憶體內計算裁剪機制的準確度損失 (Recovering Accuracy Loss of Compute-in-Memory Clipping Mechanism using Offline Error Analyzing)

利用位計數... 感知和位元向量稀疏性增強來改進位元序列快取 (Improving Bit-Serial Cache using Pop-Count-Awareness and Bit-Vector Sparsity Enhancement)more

利用位計數感知和位元向量稀疏性增強來改進位元序列快取 (Improving Bit-Serial Cache using Pop-Count-Awareness and Bit-Vector Sparsity Enhancement)

藉由行向適... 應性標記剪枝來加速注意力運算並透過剪枝補償來減少準確率損失 (Accelerating Attention Computation by Row-wise Adaptive Token Pruning and Alleviating Accuracy Drop by Pruning Compensation)more

藉由行向適應性標記剪枝來加速注意力運算並透過剪枝補償來減少準確率損失 (Accelerating Attention Computation by Row-wise Adaptive Token Pruning and Alleviating Accuracy Drop by Pruning Compensation)

階層式深度... 強化學習方法的防禦策略應用於多防禦者和多攻擊者的三維無人機reach-avoid遊戲 (A Hierarchical Deep Reinforcement Learning Approach to Defending Policies in Three-Dimensional Multi-Defender Multi-Attacker Reach-Avoid Games with UAVs)more

階層式深度強化學習方法的防禦策略應用於多防禦者和多攻擊者的三維無人機reach-avoid遊戲 (A Hierarchical Deep Reinforcement Learning Approach to Defending Policies in Three-Dimensional Multi-Defender Multi-Attacker Reach-Avoid Games with UAVs)

用於 AI... 加速器之硬體友善數字格式與基於數值感知的錯誤偵測和校正機制 (Hardware-Friendly Number Formats and Value-Aware Error Detection and Correction Mechanisms for AI Accelerators)more

用於 AI 加速器之硬體友善數字格式與基於數值感知的錯誤偵測和校正機制 (Hardware-Friendly Number Formats and Value-Aware Error Detection and Correction Mechanisms for AI Accelerators)

高效能、低... 功耗之浮點數人工智慧加速晶片架構設計 (High-Performance and Energy-Efficient Floating-Point AI Accelerator Architecture)more

高效能、低功耗之浮點數人工智慧加速晶片架構設計 (High-Performance and Energy-Efficient Floating-Point AI Accelerator Architecture)

應用於CN... N推理加速器特徵圖壓縮之編碼器與解碼器 - 設計與實作於FPGA (Encoder and Decoder for Feature Map Compression of CNN Inference Accelerator - Design and Implement on FPGA)more

應用於CNN推理加速器特徵圖壓縮之編碼器與解碼器 - 設計與實作於FPGA (Encoder and Decoder for Feature Map Compression of CNN Inference Accelerator - Design and Implement on FPGA)

利用分組式... 資料重組後之最近鄰點估算實現點雲分類任務之加速 (A Group-Based Data Rearrangement for Point Cloud Recognition's Neighbor Search Approximation)more

利用分組式資料重組後之最近鄰點估算實現點雲分類任務之加速 (A Group-Based Data Rearrangement for Point Cloud Recognition's Neighbor Search Approximation)

深度學習推... 薦模型之感知資料存取頻率修剪與量化 (Access Frequency-Aware Pruning and Quantization for Deep Learning Recommendation Model)more

深度學習推薦模型之感知資料存取頻率修剪與量化 (Access Frequency-Aware Pruning and Quantization for Deep Learning Recommendation Model)

利用減少查... 表資料技術優化神經渲染 (Optimizing Neural Rendering Using Lookup Data Reduction Techniques)more

利用減少查表資料技術優化神經渲染 (Optimizing Neural Rendering Using Lookup Data Reduction Techniques)

以指數為指... 引調整數值精度的區塊浮點數近似運算 (Block Floating-Point Approximate Computing Through Exponent-Guided Precision Adjustment)more

以指數為指引調整數值精度的區塊浮點數近似運算 (Block Floating-Point Approximate Computing Through Exponent-Guided Precision Adjustment)

不限於 R... eLU 類型深度學習模型之內積運算提早終止策略 (Early Dot-Product Termination for Computation Reduction in Deep Learning Models Supporting Non-ReLU Activation Functions)more

不限於 ReLU 類型深度學習模型之內積運算提早終止策略 (Early Dot-Product Termination for Computation Reduction in Deep Learning Models Supporting Non-ReLU Activation Functions)

利用監控脈... 波寬度調變訊號及感測地面光流增進無人機之安全性及穩定性 (Enhancing the Safety and Stability of Drones Using PWM Signal Monitoring and Ground Optical Flow Sensing)more

利用監控脈波寬度調變訊號及感測地面光流增進無人機之安全性及穩定性 (Enhancing the Safety and Stability of Drones Using PWM Signal Monitoring and Ground Optical Flow Sensing)

可程式邏輯... 閘陣列無人飛行器 (FPGA Based Unmanned Aerial Vehicle)more

可程式邏輯閘陣列無人飛行器 (FPGA Based Unmanned Aerial Vehicle)

利用卷積神... 經網路中的稀疏性改良位元序列快取記憶體 (Improving Bit-Serial Cache by Exploiting Sparsity in Convolutional Neural Networks)more

利用卷積神經網路中的稀疏性改良位元序列快取記憶體 (Improving Bit-Serial Cache by Exploiting Sparsity in Convolutional Neural Networks)

有效運用頻... 寬與儲存之可調式特徵圖壓縮式卷積神經網路推理加速器 (Bandwidth and Storage Efficient Convolution Neural Network Inference Accelerators with Adjustable Feature Map Compression)more

有效運用頻寬與儲存之可調式特徵圖壓縮式卷積神經網路推理加速器 (Bandwidth and Storage Efficient Convolution Neural Network Inference Accelerators with Adjustable Feature Map Compression)

波形可調、... 相容VGA顯示器、用於新興CMOS影像感測器之數位驅動電路矽智財 (Waveform Configurable and VGA Display Compatible Digital Driver IP for Emerging CMOS Image Sensors)more

波形可調、相容VGA顯示器、用於新興CMOS影像感測器之數位驅動電路矽智財 (Waveform Configurable and VGA Display Compatible Digital Driver IP for Emerging CMOS Image Sensors)

記憶體內運... 算之內建自測試與內建自修正策略 (Built-in Self-test and Built-in Self-repair Strategies for Computing in Memory)more

記憶體內運算之內建自測試與內建自修正策略 (Built-in Self-test and Built-in Self-repair Strategies for Computing in Memory)

基於 Ze... dboard 可程式邏輯閘 陣列的晶片測試系統 (Zedboard FPGA Based Chip Testing System)more

基於 Zedboard 可程式邏輯閘 陣列的晶片測試系統 (Zedboard FPGA Based Chip Testing System)

以硬體及演... 算法協同設計實現具位元寬度彈性之卷積神經網路及可調式非揮發記憶體內運算加速器 (Hardware-Algorithm Co-Design to Enable Bitwidth-Flexible Convolutional Neural Networks and Tunable Non-Volatile In-Memory-Computing Accelerators)more

以硬體及演算法協同設計實現具位元寬度彈性之卷積神經網路及可調式非揮發記憶體內運算加速器 (Hardware-Algorithm Co-Design to Enable Bitwidth-Flexible Convolutional Neural Networks and Tunable Non-Volatile In-Memory-Computing Accelerators)

用於快閃記... 憶體之閾值電壓感知保存錯誤修復機制 (Threshold Voltage Aware Retention Error Recovery Schemes for NAND Flash Memory)more

用於快閃記憶體之閾值電壓感知保存錯誤修復機制 (Threshold Voltage Aware Retention Error Recovery Schemes for NAND Flash Memory)

使用低位寬... 數值及低功耗記憶體進行神經網路運算 (Neural Network Computation Using Low Bitwidth Numbers and Low Power Memory)more

使用低位寬數值及低功耗記憶體進行神經網路運算 (Neural Network Computation Using Low Bitwidth Numbers and Low Power Memory)

固態硬碟內... 位址映射表之錯誤模擬、檢測、及修復 (Simulating, Detecting, and Repairing Errors in the Address Mapping Table of a Solid-State Drive)more

固態硬碟內位址映射表之錯誤模擬、檢測、及修復 (Simulating, Detecting, and Repairing Errors in the Address Mapping Table of a Solid-State Drive)

保序固態硬... 碟之設計、實現與應用 (Design, Realization, and Applications of an Order-Preserving Solid-State Drive)more

保序固態硬碟之設計、實現與應用 (Design, Realization, and Applications of an Order-Preserving Solid-State Drive)

應用於快閃... 記憶體儲存系統的JPEG圖像長期保存及修復設計 (Long-Term JPEG Data Protection and Recovery for NAND Flash-Based Solid-State Storage)more

應用於快閃記憶體儲存系統的JPEG圖像長期保存及修復設計 (Long-Term JPEG Data Protection and Recovery for NAND Flash-Based Solid-State Storage)

運用近似運... 算及省略運算之卷積神經網路低功耗設計 (Low Power Design of Convolutional Neural Networks Using Approximate Computing and Computation Skipping)more

運用近似運算及省略運算之卷積神經網路低功耗設計 (Low Power Design of Convolutional Neural Networks Using Approximate Computing and Computation Skipping)

反及閘快閃... 記憶體內JPEG圖片數據之保護及修復策略 (Protecting and Repairing Strategies forJPEG Image Data in NAND Flash Memory)more

反及閘快閃記憶體內JPEG圖片數據之保護及修復策略 (Protecting and Repairing Strategies forJPEG Image Data in NAND Flash Memory)

使用非對稱... 互連架構增進固態硬碟的讀取效能 (Increasing the Read Performance of Solid State Drives Using Desymmetrized Interconnection Architecture)more

使用非對稱互連架構增進固態硬碟的讀取效能 (Increasing the Read Performance of Solid State Drives Using Desymmetrized Interconnection Architecture)

快閃記憶體... 轉換層高速模擬器之開發及運用 (Development and Deployment of a Fast Flash Translation Layer Simulator)more

快閃記憶體轉換層高速模擬器之開發及運用 (Development and Deployment of a Fast Flash Translation Layer Simulator)