updates

content/CSE510/CSE510_L17.md

@@ -55,6 +55,27 @@ Close-loop planning:
 
 - At each state, iteratively build a search tree to evaluate actions, select the best-first action, and the move the next state.
 
+Use model as simulator to evaluate actions.
+
+#### MCTS Algorithm Overview
+
+1. Selection: Select the best-first action from the search tree
+2. Expansion: Add a new node to the search tree
+3. Simulation: Simulate the next state from the selected action
+4. Backpropagation: Update the values of the nodes in the search tree
+
+#### Policies in MCTS
+
+Tree policy:
+
+Decision policy:
+
+- Max (highest weight)
+- Robust (most visits)
+- Max-Robust (max of the two)
+
+#### Upper Confidence Bound on Trees (UCT)
+
 
 
 #### Continuous Case: Trajectory Optimization

content/CSE5313/CSE5313_L14.md

@@ -181,7 +181,7 @@ Regenerating codes, Magic #2:
 - Both decreasing functions of $d$.
 - $\Rightarrow$ Less repair-bandwidth by contacting more nodes, minimized at $d = n - 1$.
 
-### Constructing Minimum bandwidth regenerating (MBR) codes from Minimum distance codes
+### Constructing Minimum bandwidth regenerating (MBR) codes from Maximum distance separable (MDS) codes
 
 Observation: For MBR code with parameters $n, k, d$ and $\beta = 1$, one can construct MBR with parameters $n, k, d$ and any $\beta$.
 

content/CSE5313/CSE5313_L16.md

@@ -0,0 +1,62 @@
+# CSE5313 Computer Vision (Lecture 16: Exam Review)
+
+## Exam Review
+
+### Information flow graph
+
+Parameters:
+
+- $n$ is the number of nodes in the initial system (before any node leaves/crashes).
+- $k$ is the number of nodes required to reconstruct the file $k$.
+- $d$ is the number of nodes required to repair a failed node.
+- $\alpha$ is the storage at each node.
+- $\beta$ is the edge capacity **for repair**.
+- $B$ is the file size.
+
+#### Graph construction
+
+Source: System admin.
+
+Sink: Data collector.
+
+Nodes: Storage servers.
+
+Edges: Represents transmission of information. (Number of $\mathbb{F}_q$ elements is weight.)
+
+Main observation:
+
+- $k$ elements (number of servers required to reconstruct the file)  The message size is $B$. from $\mathbb{F}_q$ must "flow" from the source (system admin) to the sink (data collector).
+- Any cut $(U,\overline{U})$ which separates source from sink must have capacity at least $k$.
+
+### Bounds for local recoverable codes
+
+#### Turan's Lemma
+
+Let $G$ be a graph with $n$ vertices. Then there exists an induced directed acyclic subgraph (DAG) of $G$ on at least $\frac{n}{1+\operatorname{avg}_i(d^{out}_i)}$ nodes, where $d^{out}_i$ is the out-degree of vertex $i$.
+
+#### Bound 2
+
+Consider the induced acyclic graph $G_U$ on $U$ nodes.
+
+By the definition of $r$-locally recoverable code, each leaf node in $G_U$ must be determined by other nodes in $G\setminus G_U$, so we can safely remove all leaf nodes in $G_U$ and the remaining graph is still a DAG.
+
+Let $N\subseteq [n]\setminus U$ be the set of neighbors of $U$ in $G$.
+
+$|N|\leq r|U|\leq k-1$.
+
+Complete $n$ to be of the size $k-1$ by adding elements not in $U$.
+
+$|C_N|\leq q^{k-1}$
+
+Also $|N\cup U'|=k-1+\lfloor\frac{k-1}{r}\rfloor$
+
+All nodes in $G_U$ can be recovered from nodes in $N$.
+
+So $|C_{N\cup U'}|=|C_N|\leq q^{k-1}$.
+
+Therefore, $\max\{|I|:C_I<q^k,I\subseteq [n]\}\geq |N\cup U'|=k-1+\lfloor\frac{k-1}{r}\rfloor$.
+
+Using reduction lemma, we have $d= n-\max\{|I|:C_I<q^k,I\subseteq [n]\}\leq n-k-1-\lfloor\frac{k-1}{r}\rfloor+2=n-k-\lceil\frac{k}{r}\rceil +2$.
+
+### Reed-Solomon code
+

content/CSE5313/_meta.js

@@ -18,4 +18,5 @@ export default {
     CSE5313_L13: "CSE5313 Coding and information theory for data science (Lecture 13)",
     CSE5313_L14: "CSE5313 Coding and information theory for data science (Lecture 14)",
     CSE5313_L15: "CSE5313 Coding and information theory for data science (Lecture 15)",
+    CSE5313_L16: "CSE5313 Coding and information theory for data science (Exam Review)",
 }

content/CSE5519/CSE5519_A3.md

@@ -1,2 +1,21 @@
 # CSE5519 Advances in Computer Vision (Topic A: 2023 - 2024: Semantic Segmentation)
 
+## Segment Anything
+
+[link to the paper](https://arxiv.org/pdf/2304.02643)
+
+### Novelty in Segment Anything
+
+Brute force approach with large scale training data (400x) more
+
+#### Dataset construction
+
+- Model-assisted manual annotation
+- Semi-automatic annotation
+- Automatic annotation (predict mask for 32x32 patches)
+
+> [!TIP]
+>
+> This paper shows a remarkable breakthrough in semantic segmentation with a brute force approach using a large scale training data. The authors use a transformer encoder to get the final segmentation map.
+>
+> I'm really interested in the scalability of the model. Is there any approach to reduce the training data size or the model size with comparable performance via distillation or other techniques?